Improving Inventory and Distribution in an
Aerospace Parts and Service Organization
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
JUN 15 2011
by
LIBRARIES
Steven Allan Wessels, Jr.
Bachelor of Science in Industrial Engineering, Kettering University (2004)
Submitted to the MIT Sloan School of Management and the Engineering Systems Division in
Partial Fulfillment of the Requirements for the Degrees of
Master of Business Administration
and
Master of Science in Engineering Systems
ARGI4IVs
In conjunction with the Leaders for Global Operations Program at the
Massachusetts Institute of Technology
June 2011
© 2011 Massachusetts Institute of Technology. All rights reserved.
I--
Signature of Author
May 6, 2011
Engineering Systems Division, MIT Sloan School of Management
Certified by
Donald Rosenfield, Thesis Supervisor
Senior.k6QturerYITVfo
hool of Management
Certified by
11/s Caplice, Thesis Supervisor
tw
-Aenior
Accepted by
L turer, Engineering Systems Division
.N
Nancy Lefeson, Chair, Engineering Systems Division Education Committee
Professor, Aeronautics and Astronautics and Engineering Systems Division
Accepted by
Debbie Berechman, Exkcutive Director of MBA Program
MIT Sloan School of Management
This page intentionallyleft blank.
Improving Inventory and Distribution in an
Aerospace Parts and Service Organization
By
Steven Allan Wessels, Jr.
Submitted to the MIT Sloan School of Management and the Engineering Systems Division on
May 6, 2011 in partial fulfillment of the Requirements for the Degrees of Master of Business
Administration and Master of Science in Systems Engineering
Abstract
Hamilton Sundstrand has made several changes to their supply chain in recent years,
including increased offshore and outsourced production, new service offerings and relocation of
facilities, to meet shifting business needs to remain a top competitor in the aerospace systems
industry. This thesis reviews the distribution network of their aftermarket parts and service
business to ensure that Hamilton Sundstrand meets customer needs through efficient supply
chain design and aligning business strategy with inventory planning.
A review of the current state is employed to locate gaps in strategic design, operating
efficiencies and customer service levels. Improvement opportunities identified in the current
state analysis are addressed with proposed alternatives to adjust the distribution network to meet
current and future needs while minimizing cost and maintaining or raising service levels. The
combined proposals of relocating distribution center volumes, reducing on hand inventory at colocated sites and closing a forward stocking location are estimated to result in over one million
dollars in annual cost savings.
Thesis Supervisor: Donald Rosenfield
Title: Senior Lecturer, MIT Sloan School of Management
Thesis Supervisor: Chris Caplice
Title: Senior Lecturer, Engineering Systems Division
This page intentionally left blank.
Acknowledgments
I would like to thank the Customer Service team at Hamilton Sundstrand for being such great
hosts, partners and friends during my internship. You helped make the project successful and
provided me with a terrific learning opportunity.
Specifically I would like to thank my supervisor, Terry Gill and executive sponsor Matthew
Bromberg. Matthew provided encouragement to take a critical eye to his group's supply chain
and supported implementation of ideas that could improve it. Terry's guidance and feedback
helped me focus on the areas that needed the most attention and answered my countless
ridiculous questions. Without his support this project would not have been possible and certainly
would not have been as successful.
Thanks also to Jeff, Jason, Kirk, Andy, Jerry and the rest of the Materials Management team for
accepting me as a member of the team and giving me an education on how inventory is really
managed. A special thanks to Marie for being a great friend and an invaluable resource in
understanding how the business works and coaching me on how to pull out the right data to
visualize the problem areas.
I would like to thank my advisors, Don and Chris, for seasoned insight, guidance and valuable
feedback that directed my work to a better result.
I would like to especially thank my wife, Christina. Her love, support and motivation got me
through the six months away on internship, all while she did the majority of planning for our
fantastic wedding. I look forward to the next chapter in our lives.
Thanks to my parents and brothers, who continue to be ardent supporters and who push me to
constantly challenge myself in new endeavors.
Last, but certainly not least, I would like to thank the 2011 class of Leaders for Global
Operations. The two years at MIT would not have been the same without my great new friends.
This page intentionallyleft blank.
Table of Contents
A bstract..........................................................................................................................................
3
A cknow ledgm ents .........................................................................................................................
5
Table of Contents ..........................................................................................................................
7
List of Figures................................................................................................................................
9
1.
2.
Introduction .........................................................................................................................
1.1.
Ham ilton Sundstrand..................................................................................................
11
1.2.
Custom er Service Division.........................................................................................
14
1.3.
A chieving Competitive Excellence (A CE)................................................................
16
1.4.
Overview of Chapters...................................................................................................
17
Project O verview .................................................................................................................
19
2.1.
3.
The A fterm arket Distribution N etw ork ..................................................................
19
2.1.1.
Repair Service Sites.............................................................................................
20
2.1.2.
On Site Support (O SS) Locations.......................................................................
20
2.1.3.
Distribution Centers..............................................................................................
21
2.1.4.
Suppliers .................................................................................................................
22
2.1.5.
Transportation.......................................................................................................
22
2.2.
Problem Statem ent .........................................................................................................
23
2.3.
Project Objectives ......................................................................................................
23
2.4.
Approach........................................................................................................................
23
Current State A nalysis....................................................................................................
25
Supply Chain Strategy................................................................................................
25
3.1.
3.1.1. Supply Chain Fit to Com petitive Strategy..............................................................
3.2. Supply Chain D esign ......................................................................................................
3.2.1
Facilities..................................................................................................................
25
28
29
3.2.2.
Strategic Inventory Placem ent ...........................................................................
41
3.2.3.
Transportation M ethods.......................................................................................
44
3.3.
Supply Chain Planning..............................................................................................
45
Inventory Planning..............................................................................................
45
Chapter Sum mary.......................................................................................................
49
3.3.1
3.4.
4.
10
Alternatives Analysis.......................................................................................................
4.1.
Closing A sia Distribution Center................................................................................
4.1.1.
Estim ated Im pact ..................................................................................................
50
50
50
4.1.2.
Supporting Facts ..................................................................................................
51
4.1.3.
Opposing Facts / Alternative View s ....................................................................
53
4.1.4.
Recom m endation ..................................................................................................
54
4.2.
Relocate Rockford Distribution Center Volum e.........................................................
54
4.2.1.
Estim ated Im pact ..................................................................................................
56
4.2.2.
Supporting Facts ..................................................................................................
57
4.2.3.
Opposing Facts / Alternative View s ....................................................................
58
4.2.4.
Recom m endation .................................................................................................
59
Optimize Co-located Domestic Repair Center Inventory ...........................................
60
4.3.
4.3.1.
Repair Center Inventory Planning ......................................................................
61
4.3.2.
Reduce A verage Inventory ..................................................................................
65
4.3.3.
Optim ize A SL Stocking Policies .........................................................................
71
5.0 Summ ary ..............................................................................................................................
75
5.1
General Takeaw ays ....................................................................................................
76
5.2.
Follow -up A nalysis ....................................................................................................
77
Bibliography................................................................................................................................80
List of Figures
Figure 1 - U TC 2009 R evenues ..................................................................................................
Figure 2 - Hamilton Sundstrand Customer Service Locations .................................................
Figure 3 - Hamilton Sundstrand Customer Service Organization .............................................
Figure 4 - Simplified Network Design.......................................................................................
Figure 5 - High Level Supply Chain Flow................................................................................
Figure 6 - Cost-Responsiveness Efficient Frontier..................................................................
Figure 7 - Comparison of Efficient and Responsive Supply Chains .........................................
Figure 8 - Domestic Aftermarket Facility Locations................................................................
Figure 9 - Windsor Locks DC Inbound and Outbound Volume................................................
Figure 10 - Rockford DC Inbound and Outbound Volume .......................................................
Figure 11 - Phoenix and Puerto Rico DC Inbound and Outbound Volume .............................
Figure 12 - Distributor Storage with Carrier Delivery..............................................................
Figure 13 - Aftermarket Distribution Network Design Model ..................................................
Figure 14 - Inventory Part Matrix Organized by Facility Type................................................
Figure 15 - Inventory Part Matrix Organized by Facility Location...........................................
Figure 16 - Inventory Turns by Part Histogram.......................................................................
Figure 17 - Inventory Policies..................................................................................................
Figure 18 - Sales Demand for IATA 3 in Cost of Goods Sold ..................................................
Figure 19 - Min-Max (s, S) Continuous Review Order Policy................................................
Figure 20 - Poisson Test Part Sample.......................................................................................
Figure 21 - Windsor Locks Ordering Cost Optimization .........................................................
Figure 22 - Windsor Locks Repair Replenishment Lead Time from Local DC............
Figure 23 - Windsor Locks Repair Safety Stock Optimization Savings ..................................
Figure 24 - Revised ASL Policy Segments................................................................................
Figure 25 - Part Buckets Identified for ASL .............................................................................
12
14
16
19
21
26
27
33
34
35
36
39
40
42
43
47
48
52
61
63
67
68
69
72
73
1.
Introduction
Companies across most durable goods industries have seen a revenue growth opportunity
in extending their involvement in the product lifecycle beyond initial purchase through
replacement parts, repair services and other recurring revenue sources. The higher profitability of
services and replacement parts compared to initial production is appealing, as is the ability to
secure a long term relationship with the customer. The aerospace industry is especially engaged
in providing long term support as their product lines can see use for many years, sometimes in
excess of 20 years. Along these lines, many aircraft parts and systems suppliers have built a
strong business in aftermarket parts and service. Hamilton Sundstrand, a major aerospace parts
and systems supplier, has established itself as a leader in long term customer service. They have
been able to sustain excellent service to meet customer demand and, as a result, enabled long
term customer relationships.
Providing the high service levels demanded by their customers brings with it significant
operating costs. The Customer Service division, which oversees the aftermarket business for
Hamilton Sundstrand, meets customer demand for parts and service through a far reaching
distribution network. Fulfilling the needs of many customers worldwide with a broad product
catalog, long production lead times and quick order turnaround often means carrying substantial
inventories. The policies they set to govern inventory management and distribution logistics can
significantly impact their costs and level of service. Deciding where to place inventory, how
much inventory to keep and how to fulfill orders are all critical to balancing meeting customer
expectations and keeping costs in check.
This thesis is the result of a six-month internship project within the Customer Service
division of Hamilton Sundstrand tasked with identifying opportunities to better address the
inventory and distribution challenges of the aftermarket business and recommending alternatives
to improve cost and customer service performance. The approach used includes a combination of
supply chain strategy, single item inventory optimization, network flow, and multi-echelon
inventory analysis. These techniques are popularly used in reducing total supply chain costs and
resulted in significant savings in this analysis of Hamilton Sundstrand's aftermarket network.
1.1. Hamilton Sundstrand
Hamilton Sundstrand is a subsidiary of the United Technologies Corporation (UTC). UTC
(NYSE: UTX) "provides high technology products and services to the building systems and
aerospace industries worldwide" (UTC 10K Filing, 2009). UTC has managed significant growth
organically and through acquisitions throughout its history. In their 2009 10K filing UTC
identifies itself as having six business segments in two main categories:
o Commercial Businesses:
o Otis - elevators, escalators, moving walkways and service.
o
Carrier- heating, ventilating, air conditioning (HVAC) and refrigeration
systems, controls, services and energy efficient products for residential,
commercial, industrial and transportation applications.
o
UTC Fire & Security - fire and special hazard detection and suppression systems
and firefighting equipment, security, monitoring and rapid response systems and
service and security personnel services.
o Aerospace Businesses:
o
Pratt & Whitney - commercial, military, business jet and general aviation
aircraft engines, parts and services, industrial gas turbines, geothermal power
systems and space propulsion.
o Hamilton Sundstrand - aerospace products and aftermarket services, including
power generation, management and distribution systems, flight systems, engine
control systems, environmental control systems, fire protection and detection
systems, auxiliary power units, propeller systems and industrial products,
including air compressors, metering pumps and fluid handling equipment.
o
Sikorsky - military and commercial helicopters, aftermarket helicopter and
aircraft parts and services.
In addition to these business segments UTC operates a group called UTC Power that works on
developing and commercializing fuel cell technologies. UTC also operates a central research and
development group (Research Center) near the corporate headquarters in Hartford, CT.
5.6
UTC 2009 Revenues by Business Segment
(USD, Billions)
NOtis
0 Carrier
* UTC Fire & Security
N Pratt & Whitney
*Sikorsky
Hamilton Sundstrand
Figure 1 - UTC 2009 Revenues
UTC businesses had collective revenues of $52.9B in 2009 with 53.9% coming from the
commercial businesses and 46.1% from the aerospace group. Sixty percent of revenues were
attributed to business outside of the United States. UTC's 2009 revenues placed them as #37 on
the Fortune 500 ranking for the largest American companies for the second straight year (Fortune
Magazine, 2010). Internationally they rank #130, slipping slightly from #123 the previous year
(Fortune Magazine, 2010). As of December 31, 2009, UTC employed 206,700 personnel
worldwide, with approximately 65% of the workforce located internationally.
12
Hamilton Sundstrand represents 10.4% of the total UTC revenue for 2009 and contributes
a significant portion (24.5%) of the aerospace group. Hamilton Sundstrand is among the world's
largest suppliers of technologically advanced aerospace and industrial products. Hamilton
Sundstrand designs and manufactures aerospace systems for commercial, regional, corporate and
military aircraft, and is a major supplier for international space programs (Hamilton Sundstrand
Communications, 2010). The company has parts and systems on nearly every in-service aircraft
and has been selected as a key supplier for many new developments, including the Boeing 787
Dreamliner that will soon enter service. Hamilton Sundstrand has over 600 parts in nine different
systems on the Dreamliner, with lifetime project revenues estimated to surpass $15B. The
company is also the prime contractor for NASA's space suit and life support systems for their
international space programs.
Hamilton Sundstrand is the result of a June 10th, 1999 merger of the Sundstrand
Corporation with its Hamilton Standard business. Both Hamilton Standard and Sundstrand have
a long and prominent history in the aerospace industry. Hamilton Standard got its start in 1929
through the merger of Hamilton Aero Manufacturing and Standard Steel Propellers under
ownership of UTC's predecessor, the United Aircraft and Transport Corporation. Hamilton
Standard had propellers and other parts on such historic aircraft as Charles Lindbergh's The
Spirit of St. Louis and Amelia Earhart's Electra 1OE. Sundstrand Corporation had an earlier start
in 1905 as a tooling and milling company, gaining its name in 1926 from the founding brothers
and innovated in mechanical and electrical aerospace components throughout its history. After
the 1999 merger, Hamilton Sundstrand used its Windsor Locks, CT location as the world
headquarters. As of 2010, Hamilton Sundstrand includes 16,400 employees located at 150
facilities in 20 countries around the world (Hamilton Sundstrand Communications, 2010).
1.2. Customer Service Division
The Hamilton Sundstrand Customer Service (HSCS) division is primarily charged with
providing spare parts, repair services, training and technical support for the product lines that
Hamilton Sundstrand develops. The HSCS enterprise reaches from vendor sourcing all the way
through to sales to external customers (e.g., airlines).
-AT
IAfKTArkU--AfriAsiae
=DC
EaA
(HS)
= DC(3PL)
= DC(Subs.)
= RepairFacility
!=Onsite Support
Figure 2 - Hamilton Sundstrand Customer Service Locations
The Customer Service group supports customers through a worldwide network of facilities, with
major distribution centers in the domestic United States, repair facilities in each of the
International Air Transport Association (IATA) regions, and On Site Support (OSS) locations at
numerous customer locations. Figure 2 shows a map of the major Customer Service locations
including distribution centers (DC), repair facilities (RC) and several OSS locations. In addition
to the distribution and service infrastructure, a state-of-the-art Customer Response Center was
added in 2010 to provide customers with a single point of contact for issue resolution.
The main customer offerings of the Customer Service division are providing aftermarket
parts ("spares"), repair services and vendor managed inventory called on-site support (OSS).
Most spares have a catalog lead time to customers of seven (7) days, but in the event of an
aircraft on ground (AOG) emergency where a plane needs a critical part, turnaround times could
be required for the same day. Spares sales to external customers are generally serviced out of the
main distribution centers, as are replenishments to the repair and onsite support locations. More
recently, asset management programs have provided new opportunities for customers to "rent"
parts by the in-service time rather than buy them outright.
Repair services are provided for repairable assemblies and parts at repair centers in each
IATA region. Broken units arrive at the nearest facility associated with the product family where
they are torn down and troubleshot to determine the cause and repaired with the necessary spare
parts and adjustments. The entire repair process is typically required to occur within a 15 day
turnaround time. Spares for the repair services are stocked locally at the repair center.
OSS is a unique offering to customers that would like a vendor managed inventory solution
for their Hamilton Sundstrand parts. A long term contract is signed which provides the customer
with Hamilton Sundstrand owned and managed inventory at their own location, where the parts
are only billed once the parts are consumed. These locations have very high customer service
levels, requiring them to keep inventory for a wide variety of parts.
Among its other responsibilities, the Customer Service division manages the inventory and
transportation of spares to support their direct customer orders and replenishments for their
service locations. Ownership of this responsibility falls under the Materials Management group
within the World Wide Repair and Supply Chain organization as shown in Figure 3. Due to the
significant differences in order fulfillment and inventory controls, commercial and military
spares are managed by separate teams within the Materials Management group.
Hamilton
Sundstrand
Systems
Industrial
Aerospace I
ogsmer
World Wide
Repair and
Supply Chain
Materials
Management
Figure 3 - Hamilton Sundstrand Customer Service Organization
The senior leadership of the Customer Service group believed that there were
opportunities to reduce inventory levels within the aftermarket distribution network while still
maintaining customer service performance. The six-month internship project was initiated by
their concern in there being inefficient inventory throughout the network. The author worked
with the Materials Management group to analyze the inventory and distribution policies currently
in place and develop alternative approaches with the support of the group's subject matter
experts.
1.3. Achieving Competitive Excellence (ACE)
Hamilton Sundstrand is a performance oriented organization with a variety of metrics
from customer service to financial performance. The metrics roll up into a proprietary
performance management system, called Achieving Competitive Excellence (ACE). ACE is a
framework similar to the Toyota Production System that has been adopted across all of UTC.
ACE integrates a philosophy, a suite of well-established tools, and a commitment of total
employee engagement to increase operating efficiency, reduce waste and improve customer
satisfaction (Supplier Development, 2010).
The performance against metrics is tracked faithfully, validated with annual surveys to
customers throughout the supply chain and used as the basis for continuous improvement efforts
like kaizen events and value stream mapping. Through a rigorous, data-driven process assessed
by internal assessors, employees and their organizations progress through the qualifying, bronze,
silver and gold levels of ACE (Supplier Development, 2010). Many organizations have
continuous improvement efforts in place, but often they are strictly a set of tools with limited
executive support and become a "flavor of the month" in a string of new business initiatives.
UTC and Hamilton Sundstrand have demonstrated early success at overcoming this tendency and
integrated the ACE philosophy into all processes and more importantly their company culture.
The Customer Service group has used ACE as a framework to manage the performance
of their ever increasing customer demand responsibilities. Providing high availability of parts
and quick turnaround leads to high operating costs. ACE helps track and improve metrics like
service level, stock outs, inventory turns and total inventory investment. The ACE suite of
metrics presented a good opportunity to track past performance, establish a benchmark for the
current state of operations and validate the recommended changes implemented as a result of the
project. Many of the analysis approaches described in this thesis used the metrics as a barometer
for how operational policies and strategies were driving performance, and where opportunities
may be to improve the performance of the system.
1.4. Overview of Chapters
This introduction established a high level background of Hamilton Sundstrand and the
aftermarket business managed by the Customer Service group, the setting of the internship.
Chapter 2 will provide additional background on the specific task at hand, improving the cost
and service performance of the aftermarket service network. Chapter 3 implements many of the
tools and methods used in academics and industry to develop an understanding of the current
state of the network performance. Chapter 4 takes the findings from Chapter 3 and defines
alternatives for deeper analysis. The ultimate result of Chapter 4 is a set of recommendations
derived from the alternatives analysis that could be implemented to improve service and/or cost
performance in the aftermarket service network.
2. Project Overview
Hamilton Sundstrand's Customer Service division manages, among other things, the
aftermarket parts and repair service distribution network. The aftermarket distribution network is
essentially a spare parts supply chain reaching from component suppliers through to end
customers. As with any good supply chain design, the aftermarket supply chain is a product of its
business strategy, providing high service levels. Even with a high service strategy, it is
imperative to manage costs to maintain a profitable business. Managing that balance of service
and cost efficiency is a primary focus of the Customer Service group. This project assists the
group's ongoing efforts by analyzing the current network design and inventory policies to locate
opportunities for minimizing cost while maintaining high service levels.
2.1.
The Aftermarket Distribution Network
The aftermarket distribution network is a fairly simple design with just three main stages:
suppliers, distribution centers (DCs) and customers as shown in Figure 4. Each stage has
locations that are geographically dispersed creating a truly global supply chain.
supp&..
ocA
anstem
T
IATA I
IATA III
R
Caneer
'-On Spprt
UCr*UfiacWne
Figure 4 - Simplified Network Design
The aftermarket distribution network supplies spare parts to "external customers" including over
800 airlines, maintenance repair and overhaul (MRO) organizations as well as providing parts to
their own "internal customer" sites. The internal sites include repair service facilities and vendor
managed inventory sites for customers called on site support (OSS) locations.
2.1.1. Repair Service Sites
The repair sites receive parts or assemblies from customers that need to be refurbished
and returned within 15 days or less. Repair centers specialize in certain product families and, in
most cases, are duplicated in each market to provide quick turnaround. Markets are defined as
the three major International Air Transport Authority (IATA) regions: the Americas; Europe,
Middle East and Africa (EMEA) and Asia. In order to maintain high service levels on tight
deadlines, repair centers carry their own inventory on parts, generally several thousand unique
part numbers, required for their operations. Replenishments of parts for a repair center typically
come from only one of the four main distribution centers as they too are dedicated to certain
product families. The repair centers supporting IATA 1, the Americas, are located in the United
States, typically in the same building as or adjacent to their supplying distribution center. The
IATA 1 repair centers have the highest repair volumes and part usage of the repair network.
2.1.2. On Site Support (OSS) Locations
OSS locations are a vendor managed inventory solution that Hamilton Sundstrand
provides for customers. These locations have a limited part list, generally about 2,000 unique
parts, that is maintained locally to support the products used by that specific customer. Window
service levels, or having the part immediately available when a request is made to the stockroom,
at OSS sites are maintained above 97% and parts not immediately available must be supplied
within three to seven days. Because these sites may support multiple product families to fulfill all
of a customer's needs, they need to be replenished by more than one distribution center. OSS
locations are dispersed throughout the world depending on contract agreements with customers,
with at least one site in each IATA region.
External Supplier
Internal Supplier
Internal
Distribution /
Inventory
External
Distribution /
Inventory
Internal Customers
External Customers
HSRepr
CustomeriRepair
Receipt
M S
Figure 5 - High Level Supply Chain Flow
2.1.3. Distribution Centers
All customers, internal and external, are supplied by the distribution stage: four main
distribution centers, two forward stocking locations and a third party warehouse. The distribution
centers are all located in or near the United States; the two larger distribution centers are in
Rockford, IL and Windsor Locks, CT and two smaller sites located in Phoenix, AZ and Puerto
Rico. Parts are shipped direct from each of the distribution centers to the customer sites based on
received orders, which in the case of internal customers are replenishments for their local stock
levels. Due to the difference in customer demand lead times (under seven days) and supply lead
times (often greater than six months), distribution centers must carry inventory to buffer demand
and supply. The two forward stocking locations, located in Asia and the Middle East, were
originally set up to keep inventory of certain items in their local markets based on customer
requirements, tax advantages and speed to market. The third party warehouse keeps stock of
older parts that see very limited demand.
2.1.4. Suppliers
The supplier stage, like the customer stage, is divided into internal and external sources.
Internal sources are Hamilton Sundstrand's own manufacturing facilities that produce for
original production as well as the aftermarket. Over the past decade Hamilton Sundstrand has
transitioned from manufacturing most parts to today purchasing roughly 80% of their parts and
manufacturing the remainder. The increase in outsourced parts has created a wide base of
suppliers, however many of the outsourced parts used in the aftermarket still flow through the
Hamilton Sundstrand manufacturing facilities. Only approximately 15% of parts are sourced
directly from suppliers to aftermarket distribution centers, termed "direct receipt" parts, the
remainder is sourced from the manufacturing facility associated with that product group. Like
the IATA 1 repair centers, the domestic manufacturing facilities, or "internal suppliers," are
located at the same sites as distribution centers.
2.1.5. Transportation
Transportation within the aftermarket distribution network is almost entirely air freight,
even replenishments from suppliers to the DCs and from the DCs to the remote repair and OSS
sites. While this normally would not be cost effective for most supply chains, the nature of
aircraft parts make this transportation method feasible. Most aircraft parts are very expensive,
light weight products that fit the characteristics of air transportation. The high part costs also
drive up inventory costs, so inventory on hand is typically kept as low as possible and expedited
transportation is used to ensure service levels are maintained. Furthermore, most outbound
shipments to customers are free on board (FOB) at the shipping dock, meaning the customer
picks up the cost for transportation. As most Hamilton Sundstrand parts are run-to-failure rather
than life limited parts, a spare part is generally ordered only when needed or to refill a limited
stock and is expedited.
2.2.
Problem Statement
The high service level commitments and global reach of the aftermarket distribution
network creates a difficult inventory and supply chain challenge for the Hamilton Sundstrand
Customer Service group. One core issue is that the distribution network has a supply and demand
lead time imbalance, requiring inventory to be held to meet customer demand. The high value
aircraft parts drive up inventory cost quickly. Determining how much inventory to keep and at
which locations in the supply chain to keep them to meet high service levels at a minimized cost
is a strategic imperative for the Customer Service group. The intent of this project is to analyze
the current aftermarket distribution network for improvements in network design, inventory
policy and transportation to reduce operating costs while maintaining high service levels.
2.3.
Project Objectives
A successful project outcome provides the Customer Service leadership team with a set
of recommendations for improving their distribution network on the basis of service level
improvements and cost reductions. The recommendations provide explicit actions to implement
with respect to inventory policies, facility network design and transportation methods.
Recommendations will be supported by a cost & benefit analysis as well as impact on customer
service levels.
2.4.
Approach
The aftermarket distribution network at Hamilton Sundstrand is a mature and continuously
improving supply chain. Several enhancements have been made in recent years to increase
service and cost performance, including the implementation of a new demand forecasting system
and improvements to inventory and distribution policies. With this in mind, a hierarchical
approach to locate and rectify inefficiencies in the current system was taken rather than
undertaking a complete redesign of the supply chain. A high level analysis was performed to
identify potential sources of inefficiencies which were then refined through data analysis,
modeling and interviews with subject matter experts. Alternatives showing the greatest promise,
measured on reduced operating costs, maintained or improved service levels and impact to other
operations, were prepared as recommendations to the Customer Service leadership team for
implementation.
3.
Current State Analysis
There are many approaches to find improvements in an organization's supply chain.
When faced with a broad goal of refining the cost and service performance of an entire
distribution network, like Hamilton Sundstrand's aftermarket business, there are benefits to first
performing high level analyses to highlight problem areas and then "sharpening the pencil" in
those areas to locate the specific gains. Following this hierarchal approach, the high level
analysis of the current state begins with the overall supply chain strategy, continues on to
network planning and a cursory analysis using the logistical drivers of inventory, transportation
and facilities to identify problem areas.
3.1.
Supply Chain Strategy
It is vital to begin a cost and service analysis at the supply chain strategy level to deliver
an appropriate end solution. Many solutions can be identified that lower cost to serve, but if they
move the supply chain out of alignment with the business' competitive strategy then they
ultimately have failed to deliver value.
3.1.1. Supply Chain Fit to Competitive Strategy
A firm's competitive strategy defines, relative to its competitors, the set of customer needs that it
seeks to satisfy through its products and services (Chopra & Meindl, 2007). There are a variety
of competitive differentiators that companies typically seek to satisfy. Chopra and Meindl (2007)
describe a key strategic decision in the tradeoff between cost and responsiveness. Another
common framework for identifying customer needs segments on cost, quality, availability and
innovativeness (Hayes & Wheelwright, 1984). No matter which framework you use it is
universally understood that a firm cannot compete on all dimensions simultaneously.
Successfully delivering on one dimension requires sacrificing one or more of the other
dimensions. Chopra and Meindl (2004) represent the trade off as an efficient frontier curve,
shown in Figure 6. Providing higher levels of responsiveness comes at the expense of higher cost
as a firm moves along the curve to the upper left. This assumes, however, that the firm is
operating on the efficient frontier and many organizations are in fact inside the curve where they
can improve both cost and responsiveness by moving towards the efficient frontier.
High
C
Law
High
Lw
Cost
Figure 6 - Cost-Responsiveness Efficient Frontier
Source: Chopra and Meindl (2007)
Determining the strategic fit of the Hamilton Sundstrand aftermarket business to the costresponsiveness tradeoff is heavily impacted by its products' characteristics. For companies to be
sure that they are taking the right approach, they must first determine whether their products are
functional or innovative (Fisher, 1997). While a few of the characteristics Fisher (1997)
describes as qualifying an innovative product (i.e., short life cycle and short lead time) do not
necessarily apply to Hamilton Sundstrand's aftermarket parts, the key characteristics of uncertain
demand and high margin certainly do.
Demand uncertainty is linked to the repair use nature of aftermarket parts. The majority of
the company's parts are not life-limited, meaning they are only replaced when there are failures
or pending failures rather than on routine maintenance schedules. The parts are also very
expensive, leading to most companies carrying minimal stock to sustain their repair operations.
-
0
.
w ., ON,
II
'. -
-11-
1. -
.,
..
A - I-
I.Im- -
" I'
-
-
"
'-
These two demand characteristics drive a significant amount of uncertainty. The high margins of
the parts are typical of other aftermarket industries where initial contracts are won at thin
margins and profits are realized through service and spare parts. With these key characteristics of
high margins and uncertain demand, there is a strong strategic fit for Hamilton Sundstrand's
products with a high response supply chain.
A responsive strategy dictates the characteristics of the supply chain design. Fisher
(1997) presents a comparison of these supply chain characteristics, shown in Figure 7, that
directs the design based on factors like inventory strategy, lead time strategy and supplier
strategy. For Hamilton Sundstrand's supply chain this means a design with responsive suppliers,
higher inventories and aggressively pursuing short lead times.
Efficient Supp Chins
SUPp -nChain
Respond quickly to demand
Higher margins because price is
not a prime customer driver
Maintain capacity flexibilityto
Manufacturing
buffer against demand/supply
strategy
uncertainty
Minimize inventory to lower cost Maintain buffer inventory to deal
Inventory
with demand/supply uncertainty
strategy
Reduce aggresively, even if the
Reduce, but not at high expense
Lead time
costs are significant
strategy
Supplier strategy Select based on cost and quality Select based on speed, flexibility,
reliabilityand quality
Primary Goal
Pricing strategy
Supply demand at the lowest cost
Lower margins because price isa
prime customer driver
Lower costs through high
utilization
Figure 7 - Comparison of Efficient and Responsive Supply Chains
Adapted from Fisher (1997)
The design of the supply chain encompasses a wide breadth of topics including location
and capacity of facilities, whether to outsource, transportation modes, which markets to serve
from which locations, the inventory policies at each location and how to fulfill orders. Each of
these topics requires design decisions with impacts over varying time horizons. Chopra and
Meindl (2007) describe three categories for these decisions: supply chain design, supply chain
planning and supply chain operation. Supply chain operation decisions, like how to handle
individual customer orders, have a tightly-defined scope and a time horizon measured in days.
They are too micro-focused for an initial review. Only the supply chain design and supply chain
planning focus areas will be used in the current state analysis.
3.2. Supply Chain Design
Supply chain design is comprised of decisions with time horizons of several years. These
include whether to outsource or perform a supply chain function in-house, choosing the locations
and capacities of production and warehousing facilities, which products to be made or stored at
various locations, the modes of transportation to be made available along different shipping legs
and the type of information system to be utilized (Chopra & Meindl, 2007).
These decisions are made for the long term and are very difficult and expensive to
change. Making these decisions requires considering long term fluctuations in supply, demand,
customers and the business environment. The impacts of these decisions become operating
conditions that support or constrain the business for many years. Because of these long lasting
impacts and reluctance to revisit decisions, it is not uncommon for companies to have supply
chains that do not match their current environment. In many companies, supply chains were
designed to reflect the company's operating needs ten to twenty years earlier; it is a core source
of poor supply chain performance for many businesses (Byrnes, 2010). The antiquated supply
chain design decisions that fit earlier operating needs can be a large source of savings if
corrected to meet current needs.
Since the production and sourcing component of the aftermarket distribution network is
largely dictated by Hamilton Sundstrand's OEM activities, the focus of the supply chain design
analysis is directed at locations of and uses of aftermarket network facilities, the products stored
at each facility and the transportation methods employed.
3.2.1
Facilities
The facilities decision incorporates three main subject areas: facility role, facility location
and facility size. Each subject area of facilities decisions has a significant impact on the
capabilities and performance of the supply chain. Decisions in one area impact decisions in other
areas, for example having multiple locations to serve customers may lead to smaller location
sizes. The facilities analysis follows these subject areas for the purpose of discussion but
overlaps to discuss associations between subject areas to reflect the relationships and tradeoffs.
3.2.1.1
Facility Role
Just as a single supply chain cannot meet the market needs of cost, quality, availability,
variety or innovativeness without sacrificing in a few of the other areas, a single facility cannot
effectively or efficiently meet all the needs of the supply chain. A factory cannot perform well on
every yardstick, and simplicity and repetition breed confidence (Skinner, 1974). Beckman and
Rosenfield (2008) discuss facility focus across four main dimensions: process-focus, product- or
service-focus, market-focus and general purpose. Process-focused facilities fulfill the needs of a
particular stage of the supply chain and may be differentiated by economies of size or scope, or
technological complexity. Product-focused facilities may serve many supply chain functions, but
are devoted to a certain product or product family. Market-focused facilities serve a particular
market, typically a specific geographic area. Finally, general purpose facilities meet a variety of
the other three foci to provide flexibility.
The Hamilton Sundstrand aftermarket facilities, grouped into distribution centers, repair
centers and onsite support (OSS) locations, each have a specific role they fulfill. Distribution
centers are product focused, providing parts for specific product families. Each of the four
domestic distribution centers has a group of product families that it stocks and fulfills orders for.
Focusing on a set of product families allows for a manageable knowledge base of products
served and combining comparable products for processing, but also provides sufficient scale to
make the operation efficient. The exceptions to this rule are the two distribution facilities located
in Asia and the Middle East which are market-focused, covering a variety of product groups for
their market. Companies need to have presence in markets to compete locally, understand the
needs of that market or deal with trade barriers and local content requirements (Beckman &
Rosenfield, 2008). In the case of these two market-focused distribution centers the original
motivations were essentially local content requirements (Asia and Middle East) and tax
advantages (Asia). Since these two facilities operate counter to the typical distribution center
focus the assumptions of local content and tax advantages were tested to see if they still applied.
After reviewing regional customer contracts and current tax policies, the Middle East location
still served a market-focus, but the motivations for the Asia DC no longer applied.
Repair centers are both market-focused and product-focused. Global companies find that
they have to have separate facilities to service the larger markets (e.g., North America, Europe,
and Asia), but when they have more than one facility in a market, they focus those facilities
along product, process, or materials dimensions (Beckman & Rosenfield, 2008). This holds true
for the repair centers as each key market has multiple repair centers to meet customer demand in
the region, with each repair center in a region focused on certain product groups. Strategically it
makes sense to have repair centers market focused as they the facilities most sensitive to
responding quickly to customer demand. Repair centers must tear down, troubleshoot, repair and
return a part within 15 days. This fast transaction requires local presence in a market to maintain
high service levels. With a wide product variety it makes sense that repair centers are product
and market focused. There is sufficient volume to support multiple locations in each region and
specializing in product families confines their replenishment supply to a single distribution
center.
OSS locations are market-focused in the most extreme method, targeted to a single
customer rather than a larger market. The unique requirements of these facilities demand that
they be focused on a specific customer and support a wide variety of products. To be accurate,
the facility belongs to the customer, but houses the Hamilton Sundstrand operation. These
facilities fulfill a very niche role in serving a customer with high service levels based on
contractual requirements creating a dedicated site. The tight operating relationship provides both
access to the "market" and also effectively blocks competitor access to the market.
With exception of the Asia distribution center, the Hamilton Sundstrand aftermarket
facilities have an appropriate focus for their strategic purpose. The Asia distribution center is
reviewed in detail for impacts on inventory and transportation savings for determination of an
overall fit despite its misaligned focus.
3.2.1.2
Facility Location
Facility location is one of the most complex decisions in supply chain design. To think
through facilities location, the decision makers must understand the global nature of today's
businesses and the factors affecting them as they approach globalization (Beckman &
Rosenfield, 2008). Determining the correct location for facilities, and how many to have,
requires considering many different factors. Beckman and Rosenfield (2008) describe three
factors for considering facility locations: market access, capabilities access and low-cost access.
Market access for location parallels much of the logic of market-focus for the facility role
like access to customers and service commitments, but also includes logistic tradeoffs for
transportation costs and having a presence to understand local norms. Capabilities access drives
location decisions based on access to labor skills, technologies or a supply base with Lean or
just-in-time manufacturing capabilities that require close proximity. Finally, low-cost access
location decisions are driven by factor cost impacts like labor, energy, materials and tax
advantages.
The aftermarket distribution network facilities are mainly driven by market access, with
some crossover to capabilities and low-cost access in some sites. The domestic U.S. repair
locations benefit from being placed near original manufacturing and engineering sites for access
to capabilities of product and engineering groups. The pool of skilled labor is much larger near
the original manufacturing sites which allows for cross training, skills development and career
progression without relocation. Low-cost access drove the initial decision for the Asia
distribution center as well as the Miramar, FL repair facility. As discussed in the facility role
section, the Asia distribution center was driven by tax advantages in addition to the in-region
customer requirements. The repair facility located in Miramar services products that were
formerly supported in Rockford. Moving the operation to Florida provided advantages from
lower labor costs.
The layout of Hamilton Sundstrand's current aftermarket distribution network is a
reflection of the company's historical operations. It is driven largely by the logistical drivers of
market access; a trade off of transportation, facilities and inventory costs creating the framework
for most location decisions. As detailed in Chapter 1, Hamilton Sundstrand is the merger of
Hamilton Standard and Sundstrand. Historically those two companies focused their operations in
Windsor Locks, CT and Rockford, IL, respectively. They located their manufacturing,
distribution and repair facilities in or near the same core sites. As global business dynamics
changed they adapted by adding repair sites internationally and off-shoring production in lower
factor cost areas. Their main distribution centers stayed in Windsor Locks and Rockford, and
smaller additional distribution centers were added by their new electronics assembly sites in
Phoenix and Puerto Rico. After the merger in 1999 they continued to make network design
changes, however the distribution network still has artifacts of two supply chains grown out of
the original headquarters in Windsor Locks and Rockford.
The current state of the domestic aftermarket facility locations, as shown in Figure 8, has
all of the distribution centers located adjacent to their largest suppliers, the internal OEM sites.
Similarly, the repair centers, with the exception of the Miramar location, are also located next to
their largest suppliers, the distribution centers.
Rkockford-
S
Supplier (OEM)
Figure 8 - Domestic Aftermarket Facility Locations
This outcome is largely a product of the historical inbound and outbound traffic at each site and
the ownership of transportation costs. Hamilton Sundstrand is responsible for all transportation
costs inbound to its sites and customers are responsible for the leg of travel to their sites. Figure
9 shows the volume inbound and outbound from the Windsor Locks, CT distribution center, the
largest of the sites. Almost 80% of inbound volume (e.g., replenishments, etc.) measured in
extended costi and 74% measured in lines2 comes from the Hamilton Sundstrand original
equipment (OEM) production facility at the same location in Windsor Locks.
Customers
Suppliers
Typej
OEM
VEN
OEM
OEM
VEN
VEN
VEN
VEN
Location
Windsor Locks, CT
Italy
Puerto Rico
Phoenix
Great Britain
Connecticut
France
Pennsylvania
other
%Extd. Cost %Lines
Distribution Center
Type
79%
4%
3%
3%
2%
2%
2%
1%
4%
Windsor Locks, CF
REP
CUS
CUS
CUS
CUS
CUS
CUS
REP
CUS
CUS
CUS
74%
1%
7%
3%
0%
4%
0%
3%
5%
Location
%Extd. costI %Lines
Windsor Locks, CT
Southeast U.S.
Northeast U.S.
Midwest U.S.
Southwest U.S.
Southeast Asia
France
Netherlands
Germany
Japan
Central Asia
22%
16%
10%
7%
7%
5%
3%
3%
2%
2%
2%
21%
15%
3%
3%
9%
5%
2%
5%
2%
2%
3%
other
20%
28%
Figure 9 - Windsor Locks DC Inbound and Outbound Volume
Remaining inbound volume comes mainly from two of the other Hamilton Sundstrand
OEM facilities in Phoenix and Puerto Rico and a handful of vendors. With the inbound volume
supply source being concentrated in Windsor Locks at a transportation cost of essentially zero it
makes sense to co-locate the distribution center near the OEM site. Furthermore, when you look
at the outbound volume, a significant portion (22% of total) is provided to the Windsor Locks
repair center, a location that Hamilton Sundstrand would have to pay transportation to if the
distribution center were located elsewhere.
The distribution center in Rockford is similarly situated to reduce transportation costs
relative to inbound and outbound volumes. However, due to off-shoring production activities and
Extended cost is the product of part cost times part quantity.
2 Lines are a sub-component of orders. Each order can have one or more lines with each line detailing a request for a
quantity of a certain part number. Lines can be a reasonable representation of a part movement in the system.
34
moving Rockford repair work to Miramar there is not as strong a case for the current location as
there is for Windsor Locks. A significant portion of inbound volume, especially when viewed in
lines, now comes from Hamilton Sundstrand production in Singapore (28%), Puerto Rico (17%)
and York (16%) leaving just 29% of lines coming from Rockford OEM.
Customers
Suppliers
Type
OEM
OEM
OEM
OEM
VEN
LocatIon
%Extd. Cost %Lines
49%
21%
18%
10%
1%
1%
Rockford, IL
Singapore
Puerto Rico
York, NE
Kansas
Other
29%
28%
17%
16%
6%
4%
Distribution Center
Rockford, IL
Typel
CUS
REP
CUS
REP
CUS
CUS
REP
REP
REP
OSS
CUS
Location
Southwest U.S.
Miramar, FL
Midwest U.S.
Rockford, IL
Southeast U.S.
Northeast U.S.
Singapore
Ireland
France
Germany
Central Asia
Other
%Extd. Cost %Lnes
15%
15%
8%
7%
7%
7%
4%
3%
3%
3%
3%
25%
6%
17%
2%
5%
7%
3%
6%
3%
6%
5%
6%
33%
Figure 10 - Rockford DC Inbound and Outbound Volume
The outbound volume to Rockford repair is now only approximately five percent of Rockford
DC outbound lines, being surpassed by Miramar which now makes up 17% of lines. Other repair
destinations that Hamilton Sundstrand must pay transportation for are located overseas and have
approximately the same shipping rates based on international zones 3 regardless of origin within
the United States. With these facts under consideration, the Rockford DC becomes a candidate
for possible relocation or consolidation with other sites. Transportation costs are only one
component of logistics, however, and further consideration is necessary to consider impacts on
inventory levels, customer service and facility capabilities. These topics are discussed in more
detail in the strategic planning section.
The two smaller distribution sites, Phoenix and Puerto Rico, are also co-located with their
manufacturing sites and repair centers. Figure 11 displays the inbound and outbound volume for
3FedEx Rate and Service guides were studied from a variety of origins to determine transportation advantages to
international destinations.
the two sites. Both Puerto Rico and Phoenix are even more dependent on their local OEM
supplier base, essentially being a finished goods inventory stock room dedicated to aftermarket
sales for their site. Similar to Rockford and Windsor Locks, the only non-external, large
customers are the repair centers located on site.
Type
Location
Suppliers
%Extd. costI %Lines
Distribution Center
Type
CUS
REP
CUS
CUS
OEM Phoenix
OEM Puerto Rico
other
97%
3%
<1%
75%
24%
1%
Phoenix
OEM Puerto Rico
Other
98%
2%
99%
1%
Puerto Rico
Customers
Extd. Cost %Lines
Location
%
Northeast U.S.
Phoenix
Southwest U.S.
Southeast U.S.
Other
41%
20%
12%
11%
16%
8%
48%
9%
7%
29%
CUS Midwest U.S.
CUS Southeast U.S.
REP Puerto Rico
CUS Southeast U.S.
Other
25%
22%
16%
13%
24%
15%
26%
15%
11%
33%
Figure 11 - Phoenix and Puerto Rico DC Inbound and Outbound Volume
The remaining sites in the aftermarket distribution network, the Middle East forward
stocking location, the OSS locations and the international repair centers in IATA regions 2 and 3
are all located based on drivers of market access and service commitments. OSS locations are
located explicitly on customer sites. The Middle East forward stocking location is similarly
located to serve certain customers in the region with a limited part list. The international repair
centers provide services for their associated product groups in the region to maintain quick
turnarounds. Throughout the network, all sites are located near airport facilities enabling easy
access to air freight for transportation.
3.2.1.3
Facility Size
Referencing the size of a facility typically refers to its production capacity rather than its
physical size. In terms of distribution or warehousing the size or capacity generally refers to the
holding capacity of the building. In developing a facilities strategy, it is critical to consider (a)
ways in which volumes may be built and then leveraged globally and (b) the economies of scale
associated with breaking up that volume among facilities (Beckman & Rosenfield, 2008). Sizing
a facility is a tradeoff between storage requirements and minimizing fixed costs per unit. While
inventory levels fluctuate over the short term, facility size remains constant and the cost of
maintaining the space must be covered regardless of how much space is used. In most cases the
fixed costs related to a facility size are step functions where facility size and the related costs can
only be adjusted in larger increments by adding or removing capacity.
Hamilton Sundstrand facilities, in contrast, benefit from economies in scope with mixed
use sites. Sharing distribution sites with manufacturing and repair facilities allows distribution
space allocated to aftermarket parts to be more variable. Within the limits of the larger buildings
and other business needs, they can adjust the space required for aftermarket distribution space
based on changing inventory levels. The forward stocking locations also benefit from a more
flexible sizing option as they use contracted space from third party logistics (3PL) providers. The
domestic repair centers, being co-located with the production and distribution locations, also
have flexible capacity. OSS location sizes are determined by the terms of the customer
agreement and use the customer's defined space.
The remaining facilities for sizing consideration are the international repair sites. While
these sites may benefit from relative adjustments in size or layout to benefit costs locally, they do
not lend themselves to larger economies of scale consolidation. Each site within a region handles
unique product groups and has little to gain from consolidating with another repair center in the
area.
3.2.1.4
Network Design
A distribution network is more than just a collection of facilities of various roles,
locations and sizes. Each facility fulfills a function as part of the larger system, creating a
network of paths and locations through which products are delivered to the customers. Firms can
make many different choices when designing their distribution network. The appropriate choice
of distribution network results in customer needs being satisfied at the lowest possible cost
(Chopra & Meindl, 2007). Design decisions for distribution networks balance number of
facilities and network design with response time and operating cost.
Typically, a higher response distribution network requires more facilities, located close to
the customer. Adding facilities comes at the expense of higher facility and inventory costs but
saves on transportation cost. In Hamilton Sundstrand's aftermarket network, however, the use of
air freight provides a quick response time between the distribution center and the customer
without being located nearby. Adding facilities would only marginally increase response time,
while duplicating costly inventory storage locations and increasing facility costs. Additionally,
the high priced parts lead to high inventory costs at each location, further motivating fewer
inventory stocking locations.
The design of the part flow through the system also impacts response and operating costs.
There are many distribution network designs including direct shipping, cross dock operations
and various distribution center approaches; however the characteristics of the Hamilton
Sundstrand aftermarket narrow the field to just a few appropriate models. Due to the uncertainty
of supply and demand, inventory must be kept somewhere in the system. Not having a retail
presence leaves manufacturing sites or distribution centers as options. While the Hamilton
Sundstrand manufacturing sites do have finished goods inventories, they are typically reserved
for original production inventory destined for the large air framers, not for the aftermarket. The
aftermarket network's need for dedicated parts and reliance on receiving parts from multiple
suppliers supports consolidation of inventories into a middle tier of dedicated distribution
centers.
The final design choice is how the parts will flow from the distribution center to the
customer. With a wide array of customers across the globe, maintaining a delivery fleet would be
cost prohibitive, so final mile delivery by a common carrier is appropriate. Chopra and Meindl
(2007) discuss a distribution model similar to these characteristics called distributor storage with
carrier delivery, shown in Figure 12. This model provides fast response time, better customer
experience and better order visibility than other models but coming at the cost of higher
inventory levels.
Factories
Warehouse
Storage
Customers
Figure 12 - Distributor Storage with Carrier Delivery
Adapted from Chopra and Meindl (2007)
To validate the network design used in the current state a model was built by the author
to show the actual flow of parts in the aftermarket distribution system. Each location in the
network was added to a Microsoft Excel model in a network layout shown in Figure 13. The
model was laid out similar to the distributor storage model described by Chopra and Meindl
39
(2007), with suppliers on the left, customers on the right and distribution centers in the middle.
Locations were further organized by geography to give some representation of regional part
flow. Sales and inventory transactions data were aggregated and formulated to show point to
point part movements. The aggregated sales and inventory data was added to the model by
drawing arcs between the supplying node and the receiving node. The resulting model shows real
part flow demonstrating the actual, implemented network design.
Figure 13 -Aftermarket Distribution Network Design Model
The model closely matches the distributor model described by Chopra and Meindl (2007)
as a good fit for high responsiveness, but also correspondingly higher inventory costs. Further
analysis of the model reinforced the limited role played by the Asia forward stocking location in
servicing the customers in its region. Rockford and Windsor Locks distribution centers are
reinforced as the primary distribution centers, having high volumes and serving wide customer
bases, yet supplied by a minimal set of suppliers.
3.2.2.
Strategic Inventory Placement
Inventory serves a unique purpose in buffering the uncertainty in supply and demand. It
also requires a significant investment in capital that could be used elsewhere in the business to
create value. Locating inventory in the distribution network effectively can meet customer
demands while minimizing the capital investment. Because inventory offsets supply and demand
imbalance it is best located at the push-pull boundary where customer orders initiate the "pull" of
parts to fulfill orders and the upstream supply chain "pushes" product to that point based on a
forecast. The push-pull boundary is the point along the supply chain time line where there is a
need to coordinate the two supply chain strategies, typically through buffer inventory (SimchiLevi, Simchi-Levi, & Kaminsky, 2003).
In the aftermarket distribution network the push-pull boundary occurs at the distribution
centers. Up through that point the parts are moved based on forecasts and after that point parts
are moved based on customer demand or replenishments to repair centers and OSS locations. For
that reason it would make sense that the majority of the inventory be placed at the distribution
centers. Inventory is also necessary in a multi-echelon system where downstream sites, like
repair and OSS locations, have demand lead time shorter than supply lead time. With each
distribution center being dedicated to particular product groups it would be expected that very
little inventory would be held in common at the DC tier, but with each repair center duplicated in
each region and OSS location serving multiple product groups, inventory would likely be
duplicated at the downstream tiers.
An analysis of inventory was performed on the aftermarket distribution network to
determine where stock was held for each part and if the locations matched the expected outcome.
Data was pulled from the Hamilton Sundstrand enterprise resource planning (ERP) software for
each part number and location pairing and aggregated in a matrix to visualize part duplication.
The part matrix is arranged with the same facilities listed on the rows and columns in the same
order and where they intersect the percentage of parts in common between those two sites is
shown. The matrix, shown in Figure 14, is conditionally formatted to show darker colors where
greater percentages of parts in common are held (e.g., >10% yellow, >60% orange, etc.).
8
e
D
0
DC-Rokfor
DC-Wionsrb
DC -RAsa r
id
DC - Moe
c
c o
c
x
0
0
0
0
o
01.20
1-
ox
at
2x
.
.3
mxa
0
x
s-
x
0
0
0
0
sa
x
0
a
0
0
Ox
ox Os- o
x4fl
0
ex
x
10
0
x 0.
0'
ox
Ox e x
1x e
O
0 0
x
ex 0ax
ex
.
x
0
x
20
x
x
4x
Pxf
-n
OSC
o~c
Rss-Rockfd
Os51
P- Wndoocsox
E
Ox
-
1
ox
3-
Milamsta
REP-SS a
te
is0exs
-3*
ox ox
4%Nnc Vtx
4x
s
ox P i
14 x Inventox
Fuere
x
aasi
a
Expctd,
a
0
OeX0
3- Vadili O
6%in
ex drt
par eld6x
4Y ex
REP -- S nare
1
REP - Sihoapore 2
2% 0x
ON 0.e
x
Y.
x
IN
x
3Y
x
Px x
3% 0Ao
N
x
e10
xf
x
x
ax
I
et
ox Vix
10x
N. N
l
ex
X
sa
arsu
t.
o
01
1
K
ex
os
t or 1
5ex , r/
0
0
ox 11
m
5%
0
'
x OA
10K
X tx
x
V
5e
U
4% V90xax
ou
ex
1
ft
sx
1 2
aixyex
1x
s
0
1x
4Y.0
e
X
f
%W
words, a particular part is only kept in one of the main distribution centers. This is visible in the
upper left comrer of the diagram. The exceptions, or the dark areas off of the diagonal, are
42
t
1%
a
ax
DYx
Ox
t W Nf
In 3A
other
the
I% ies
N 4x &%
ebexw
1X 2 7% 1XX Ix
Mx M
-
5xA
ft
1x
by Fa0% l
Oaned
x 2x
A. -
o
A
Y
a
ex
x
r/.
CA
oxoxe
Ix 1%exy
oW
W
ox
1
2
enex
ex
4
o
faex
t
a4
arx
iterx
1x -m
01
0
oxaxtIN
0%2
0%ooa
y2
0
ox
2x
2 x a/
xio2
o x
0
x
31 ex %
REP7 -.
REP9
40
ckfor
o
REP -
0
V.x
o
0
0S20
Wx
ox
x
x
1xoni
DSC -x
ix
1x
0 x
ex x 0 a4020
x
o0
xo5
ox
ox ex 7x
ix
00
1x
B
x
a
ox
axo
2x
12
'a
fX 12
ex
am
aex
.
2
attributed to the two forward stocking locations in Asia and the Middle East. This is intuitive as
these distribution locations are intended to have duplicate inventory to serve the local market.
However, as determined in earlier analysis, the market-access case for the Asia distribution
location was not substantiated, so this duplicate inventory may provide a savings opportunity
without sacrificing service. Continuing through the part matrix to the OSS cross section, the very
middle area, the expected high degree of part duplication is seen. The repair center cross section,
in the lower right corner, also displays the expected duplication. The repair inventory duplication
is not as common as in OSS sites, but is clustered around repair centers that support the same
product groups.
The part matrix organized by facility type matches the expected outcome, showing stock in
the locations where demand outpaces supply. What seems counterintuitive is the amount of stock
at repair facilities that are co-located with their supply source distribution centers. The
replenishment lead times at these locations should be short enough to dictate little to no
inventory. Rearranging the part matrix to be sorted by facility location and limiting the facilities
included to only co-located sites, as shown in Figure 15, magnifies this issue.
D
a. -.
REP - Phonix.
DC
Pero
Rco
o10
,a
aU UJ
1x
a
0~
0
2
or
j
us
oa
Im
.
ex
14
a2xe cc
Dc -Phoenix
Nx N~ U
ft
REP -Phoenix
1% N
01% Ox 14x
A
DC - Puerto Rico
REP -Puerto Rico
4%
Dc -Rockford
REP - Rockford
' t
ix3-
DC - Wndsor Locks
1%
REP - Windsor Locks
1x
?,
O
ft
2/.9
a
51/
31/
71
rixx
Sx3.
Dx
3-
2w-
4%
Tx
0
Ox 0-
1%
O-
x
D-
Figure 15 - Inventory Part Matrix Organized by Facility Location
The duplicate inventory between distribution centers and repair facilities at these four physical
locations presents a large opportunity for cost reduction. Approximately 20% of network
inventory is located at these four repair sites. Across these four locations the total duplicate
inventory accounts for almost $4 million dollars. Refining inventory policies at those locations
may save a significant amount of that inventory investment without impacting service levels.
Further analysis into the inventory policies at each facility is reviewed in the supply chain
planning section.
3.2.3.
Transportation Methods
Managing distribution logistics in any industry requires a balance of inventory and
transportation costs, but the characteristics of the products managed can heavily impact the
optimal policies. Cheaper, high volume products like coal are best transported via slow, cheap
methods (e.g., train or ocean freight) and stored in large inventories. High value, low volume
goods like electronics may be transported expeditiously (e.g., air freight) and kept with minimal
inventories. In the aerospace industry, the majority of parts fall into that latter group. Most parts
used in aircraft are crafted of specialized materials, comprised of proprietary electronics or
sophisticated mechanical systems that all must fit in a relatively light and small package. These
characteristics typically result in very expensive, light weight parts that lend themselves to
minimal inventory stores and rapid transportation methods.
The low weight, high value motivation for air freight is further amplified by the
replacement use nature of aircraft spare parts, especially non-life limited parts. Many parts on
airplanes are life limited, requiring that they be replaced based on some pre-determined
maintenance schedule. These parts are easier to predict and schedule replacement parts for which
reduces uncertainty to some extent and allows for slower transit modes. Most of Hamilton
Sundstrand's parts are not life limited and are replaced on less regular intervals. These parts are
more difficult to forecast needs for and the airline customers typically need the parts expedited to
replace a broken part or replace minimal on hand inventories. Even in non-critical orders the cost
of air freight is a small fraction of the cost of the part as a percent of the total transaction.
3.3.
Supply Chain Planning
Chopra and Meindl (2007) describe supply chain planning as the set of decisions
regarding which markets will be served by which locations, the subcontracting of operations, the
inventory policies to be followed and the timing and size of marketing and price promotions.
These decisions apply to a tighter time horizon than supply chain design, generally on a quarterly
or annual basis. As most Hamilton Sundstrand aftermarket locations have specified markets,
limited promotions and minimal sub contracting, the focus of the planning decisions for the
current state analysis is on their inventory planning methods.
3.3.1
Inventory Planning
As discussed in the strategic inventory placement section, Hamilton Sundstrand has
located inventory in the preferred locations to buffer supply and demand uncertainty. However,
placing inventory in the right locations is only one component of a successful inventory
investment. Poor inventory management can lead to having too much inventory on parts that are
not needed and not enough for parts in high demand. It may make sense to keep fast moving
parts in distribution centers, repair centers and OSS locations to meet demand, but slow moving
parts located only in distribution centers to minimize inventory expense. To keep inventory at
ideal levels, inventory policies are used to meet service performance goals while controlling
costs.
The service performance in the aftermarket is measured with two key metrics, customer
service level (CSL) and "stocked out" percentage. Customer service level is measured in line fill
rate, or the percent of order lines that are filled on time, a common distribution performance
measure. Silver Pyke and Peterson (1998) describe fill rate as the fraction of customer demand
that is met routinely; that is, without backorders or lost sales. Fill rate service level goals vary
based on the type of customer. For most customers a goal of 95% is targeted for lines filled
within the demand lead time and for OSS locations the goal is 97% or higher. "Stocked out"
performance measures the percent of parts during a reporting period that do not have inventory
on hand to fulfill orders. The target for "stocked out" parts is to remain below 2.5% of total parts
at distribution centers and 1.0% at repair centers. Hamilton Sundstrand has had strong
performance in meeting or exceeding their service level performance measures both in aggregate
and at individual sites. While "stocked out" performance is a useful measure for reporting overall
part inventory performance, it is not the key statistic used to set inventory levels. Inventory
policy is set by trying to meet the fill rate service level performance.
Inventory cost performance is measured in aggregate by total cost and the inventory turns
metric. Total cost of inventory is limiting as a performance benchmark as it does not relate
directly to changes in number of products or demand. Inventory turns is more directional by
comparing average inventory to actual sales and, as such, attracts more attention as a cost
measure. The inventory turns goal fluctuates, but generally is close to 4.0. The Customer Service
materials management team has also been successful in coming close to or meeting this goal in
most months as well.
While the inventory turns metric is helpful, it hides the problem of slow moving parts by
averaging turnover performance across large product families (typically the lowest visibility into
performance). These parts drive cost up while not supporting service levels. For example, out of
4,515 active parts in one repair facility the average inventory turns performance was 3.55 but had
a mean of 1.84. The distribution of those parts, displayed as a histogram in Figure 16, showed
that 67% of the parts had turns lower than the mean of 3.55 but were compensated for in the
metric by a minority of parts with high turns. Refining inventory policies can help minimize the
inventory of slow moving parts and raise the overall performance of the site.
Inventory Turns Histogram
100%
70%
50%
40%
30%20% 10% <=1
<=2 <=3 <=4
<=5 <=6 <=7
<=8 <=9 <=10 >10
Inventory Turns
Figure 16 - Inventory Turns by Part Histogram
To minimize the cost of inventory, like reducing the slow moving parts discussed above,
while also meeting service levels the Customer Service team maintains a broad set of inventory
policies. The policies dictate which parts are authorized to be stocked based on demand, called
the Authorized Stock List (ASL), and what the service levels and order quantities should be for
those on the ASL. For parts that have insufficient demand to not be placed on the ASL they have
their inventory targets set to 0. The requirements to qualify for the ASL and the service levels
and order quantities for ASL parts differ based on facility type (i.e., distribution center or repair
center) and part cost. Figure 17 shows a representation of the current qualifiers for ASL and the
associated inventory parameters for ASL parts.
Customer Service Level FillRate)
Events Per Year
6-19
28t0
2-5
Location PartCost
99%
99%
99%
$0- 10
DC
85%
95%
$10 -100
85%
85%
93%
$100 -400
85%
93%
80%
80%
$400 -1,000
93%
75%
75%
$1000 -4,000
$4,000+
RC
$0-10
$10-100
$10 -400
$400-1,000
$1,000 - 4,000
$4,00+
Months of Supply
Location Part Cost
DC
$0 -10
$10 -100
$100 -400
$400 - 1,000
$1,000 -4,00
Events Per Yer
20+
2-5
6.00
6.00
6.00
2.50
2.50
2.50
1.50
1.50
1.50
1.50
1.50
L50
0.50
0.50
0.50
stocking Policy
Location
OR ASLKi
DC
>=-2 order evennts/year
RC
Part cost >=$L )00: >= 6 order events/year
Part cost <$1,0 0: >=-2 order events/year
Ordering Cost
cost
Location
$4,000+
0.50
0.50
0.50
DC
$60
4.50
1.13
L13
4.50
1.13
L13
4.50
1.13
1.13
RC
$15
L13
1.13
1.13
90%
$0-10
$10-100
$10-400
$400-N100
$1,0 -4,0u
0.25
0.25
0.25
9
$4,000+
0.25
0.25
0.25
75%
75%
93%
100%
97%
97%
100%
97%
97%
100%
97%
97%j
97%
90%
97%
97%
90%
90%
90%
RC
Figure 17 - Inventory Policies
The inventory policies are implemented in the planning tool, Servigistics, which uses
them to administer inventory levels, trigger orders and manage demand. As mentioned
previously, parts that are identified to be on the ASL get their ordering policies set to allow for
replenishment based on the dictated service levels, ordering cost and carrying cost (20%). Order
quantities are determined by the lesser of economic order quantity (EOQ) or months of supply
maximum. The policies, while specialized to facility roles (i.e., DC or RC), are not customized to
each site. For instance, the ordering cost, which drives the EOQ calculation, is set to $15 for all
repair locations regardless of supply source. A repair facility in Europe that is supplied by
Windsor Locks DC and requires transportation has the same $15 order cost as the Windsor
Locks repair center located in the same building as the DC. The $15 amount had been reduced
from its original value of $35; however this value is still arbitrary with remaining opportunity for
refinement. Similar to the ordering cost, the ASL qualifier policies are also generic to facility
roles. The European repair center likely should have more parts on ASL than the Windsor Locks
repair center to account for the different replenishment abilities. Refining the ASL policies,
service levels and inventory parameters should aid in reducing the $4 million in duplicate
inventory identified in the strategic inventory placement analysis.
3.4.
Chapter Summary
Reviewing the aftermarket distribution network in terms of supply chain design and
supply chain planning decisions helped focus the improvement efforts on a few areas of
opportunity. While much of the aftermarket network aligns with meeting customer service needs
in a responsive, cost-effective way there are a few opportunities for refinement that require
further analysis:
e
Asia distribution center: the Asia distribution center has high duplicate
inventory and does not appear to be necessary based on market-access or
responsiveness needs.
" Rockford distribution center: the off shoring of manufacturing and relocation of
former Rockford repair work weakens the case for the Rockford distribution site.
" High duplicate inventory in co-located distribution center and repair
facilities: while inventory is necessary to buffer supply and demand uncertainty,
the repair centers located adjacent to domestic distribution centers appear to have
a surplus amount.
e
Generalized inventory policies: the policies that drive order replenishment,
safety stocks and stocking policy are non-specific to each facility, despite
proximity to supply source.
In the next chapter alternatives are generated and further analyzed to address these
opportunities with the intent of realigning the design and planning of the aftermarket distribution
network with current operating conditions.
4.
Alternatives Analysis
This chapter builds on the observations made in the current state analysis by providing
alternatives that intend to improve the Hamilton Sundstrand aftermarket distribution network.
Each alternative is analyzed for the estimated impact in reducing annual operating costs as well
as the impact on the design and planning of the distribution network. Other benefits and
drawbacks are presented in addition to the financial impact to discuss the broader issues of
making the proposed changes. Finally, each alternative is accompanied by a final
recommendation with a proposed implementation method.
4.1.
Closing Asia Distribution Center
The Hamilton Sundstrand aftermarket parts distribution network is chiefly serviced by
four main distribution centers (DCs) located in Windsor Locks, Rockford, Phoenix and Puerto
Rico. The practice of the industry to fulfill orders and replenishments by air transportation allows
for adequate servicing of global demand from a centralized distribution source. As such, the
distribution network has been maintained so that each unique part is only serviced from one
distribution center, specific to its product family. However, over time two additional sub-DCs
have been added, in Asia and the Middle East, to meet customer service commitments. These
two sub-DCs are the only locations that operate as forward stocking locations or otherwise stock
a significant amount of parts that are carried at one of the other DCs. The analysis presented in
this section reviews the business case for the continued use of the Asia DC location.
4.1.1. Estimated Impact
The savings from closing the Asia facility is derived from two main categories, the
annual operating lease with the third party logistics (3PL) provider and the carrying cost of the
inventory kept at the facility. The 3PL lease totals approximately $250,000 per year; however,
the 3PL distribution contract actually includes services for Hamilton Sundstrand beyond the
aftermarket needs, so the full amount would not be recoverable. It is estimated that half of the
services in the contract, or $125,000/year could be reduced by removing the aftermarket services.
The inventory at the Asia DC is currently valued at $540,000 at domestic standard costing rates.
Closing a DC and repositioning inventory to another DC is generally not a true cost savings as it
just moves expense from one location to another. With the Asia DC, however, the relatively high
inventory on hand is due to initial provisioning, not to support regular customer demand. The
Asia DC parts have been largely sitting in inventory without servicing demand in the region. An
analysis of current demand on Asia DC parts showed that only $70K of current inventory would
become long term inventory at other network locations, with the majority of the remainder able
to be depleted in 6-12 months after relocation. The short term inventory of $470,000 has an
annual holding cost of $94,000 using the corporate carrying cost rate of 20%. The combined
savings of reduction in the annual 3PL service contract and inventory cost bum down is
$219,000 per year. There would be additional costs savings realized from transportation to Asia
DC in the form of existing replenishment activity, however due to low demand at the site this
expense and the related savings were estimated to be insignificant.
4.1.2. Supporting Facts
The business case for opening the Asia DC originated from existing assumptions of customer
requirements for in-region parts, tax incentives for routing parts through Asia and the ability to
provide rapid turnaround for IATA III customers. An analysis of these factors under current
conditions showed that each of these benefits or requirements have limited impact.
*
Customer agreements do not require in-region parts: A review of existing product
service agreements does not strictly require the parts to be fulfilled from within the
region. As long as customer service levels can be maintained from other sources the
customer requirements are met.
*
Tax benefits no longer require shipment origin from Asia: Originally it was
understood that parts had to be shipped from within the region to get associated tax
benefits. However, in recent years new information and/or policies have come available
that attribute tax benefits simply to arrival and consumption in the region rather than
origination.
IIATA 3 market demand can be fully met by domestic distribution centers: The Asia
DC is currently only servicing 5% of IATA 3 demand, with the remaining 95% of
demand being fulfilled by the domestic DCs 4 . The existing Asia demand can easily be
absorbed by the other core distribution centers while still meeting customer service
levels. There are no parts unique to the Asia DC; each part has a duplicate source in the
U.S. Furthermore, in many cases external customers will benefit from consolidated orders
shipping from a single site.
IATA 3 COGS
U SING
0 Other DC's
Figure 18 - Sales Demand for IATA 3 in Cost of Goods Sold
4 Based on 2009 sales figures sourced from the Hamilton Sundstrand enterprise resource planning tool J. D. Edwards
(JDE).
4.1.3. Opposing Facts / Alternative Views
While there are significant benefits to closing the Asia distribution center there are also
drawbacks and alternatives that were evaluated during the analysis.
" Inventory policies could be adjusted to fulfill more demand from Asia DC: The
current five percent of sales to IATA 3 that originates from Asia could be increased by
adjusting sourcing policies in JDE and order administration. While more demand could
be driven through Asia DC, it still would not create significant value over shipping from
domestic locations.
*
Customers may push back on increased freight charges: One disadvantage to the
external customer is that the few orders that do originate from Asia today would now ship
from an origin farther away, increasing the transportation cost that they are accountable
for. The increase in freight charges would be marginal and customs fees would be similar
to what they experience from Asia today.
" Long term Asia market strategy may require a regional distribution center in the
future: With the increase in passenger miles, addition of new airlines and emerging air
framers in Asia there may be a need for a distribution center in the region in the coming
years. Even with probable customer demand and likely new service agreements in the
region, the appropriate location of an Asia regional distribution center is not yet certain.
Additionally, using third party vendors for distribution lowers switching costs attributed
to leaving now and potentially returning later are outweighed by the savings achieved by
closing the location.
4.1.4. Recommendation
After a thorough review of the facts relevant to the continued operation of the Asia
Distribution Center, the author recommends closing the facility, or in this case removing the
aftermarket parts from the 3PL contract, and serving the demand from the domestic DC network.
As the Asia distribution center is outsourced on an annual contract, exiting from the
arrangement should be done at the next annual contract renewal or prior to that if there are no
early termination fees associated. The inventory at the Asia DC should be redistributed to the
appropriate domestic distribution centers in consolidated shipments as soon as possible to
expedite the burn down in excess network inventory. Prior to relocating the inventory, customers
must be notified of the change in the service network and the order administration system must
be updated to remove the option of shipping parts from the Asia distribution center.
4.2.
Relocate Rockford Distribution Center Volume
The key drivers in the location of the distribution centers in the aftermarket network are
the reduction of transportation costs, primarily on inbound, the use of existing facilities at the
OEM and repair site, and the use of air freight to reach global demand of customers regardless of
location. A high mix of parts supplied from the local OEM and relatively high mix of internal
customer demand going to the local repair site made co-location with OEM and repair a perfect
placement for most of the core domestic distribution centers. The Rockford DC, however,
showed only 29% of inbound lines coming from the local supplier (Rockford OEM) and only 5%
of outbound lines going to the local Rockford repair. Recent changes in the business had
increased sourcing of parts internationally and a significant portion of the Rockford repair
business had moved to Florida, changing the key parameters of the Rockford location equation.
In an effort to realign the aftermarket distribution network with current business conditions a
number of alternatives were considered.
First, an idea was considered that followed the recent industry preference for placing a
central mega-dc in a low cost area close to major package carrier locations like FedEx in
Memphis, TN and UPS in Louisville, KY. This option works for many organizations because
they centralize multiple regional stocks of the same parts and benefit from consolidated inbound
shipments (Andreoli, Goodchild, & Vitasek, 2010). In the Hamilton Sundstrand business these
major gains do not apply. The four distribution centers carry different part families that would
not benefit from further consolidation and moving them farther from their source OEMs would
only increase transportation expense while adding increased facility expenses of a new site.
The second option considered, realizing there would not be significant gains in inventory,
was to consolidate the Rockford DC with the Windsor Locks location. A thorough analysis of
transportation costs, inventory savings and opportunity for consolidated outbound shipments was
prepared. The outcome was that any gains in inventory and consolidated shipping were offset
from the added transportation of product still sourced from Rockford OEM and consumed in
Rockford repair. The consolidated outbound shipping was minimal as repair sites tied to only
one distribution center leaving only small volume OSS centers that benefited. Furthermore,
consolidating the two largest DCs added significant additional workload for outbound shipping
dock in Windsor Locks and reduced flexibility in their distribution network if operations in
Windsor Locks halted due to work stoppage, weather, equipment failure or some other event.
The final alternative considered was to relocate portions of the Rockford DC volume to
better align with the change in sources and uses. Rather than completely relocate the DC, the
parts that serviced mainly the Miramar, FL repair center would be moved to that location, the
parts that did benefit from consolidation in Windsor Locks would be moved to that DC and the
remaining parts would stay in Rockford. This alternative appeared to make the most sense based
on initial calculations and was pursued for further analysis to consider it as a final
recommendation.
4.2.1.
Estimated Impact
The Rockford DC split alternative benefits the network by rebalancing the inventory to
locations with existing facilities, minimizing unneeded transportation costs and allowing for the
application of the proposed repair-related inventory reduction methods to be applied in Miramar,
a large repair facility. In addition, relocating the DC inventory would also change the required
staffing at each site; a shift in staffing to the lower cost facility in Miramar would have
associated labor savings.
To begin the analysis the demand for each part family was reviewed to see what percent
of internal customer demand was used at each location or its subordinates to determine its best fit
with the three locations. The demand-driven split of the part families resulted in 76% of parts
being allocated to Miramar, 8% to Windsor locks and 16% staying in Rockford. With each of the
proposed changes in part family locations the impact on transportation costs and inventory
reduction opportunity was tracked.
Over 25% of parts moved to Miramar and Windsor Locks were internationally-sourced
and had no cost difference from intra-U.S. destination changes. The remaining inbound volume
had to be reviewed for zone changes and associated cost differences per line. The resulting
impact was an increase in freight costs of $30,600 per year for inbound shipments. On the
outbound side the savings on replenishments to Miramar repair and consolidated shipping out of
Windsor Locks to OSS sites resulted in a decrease in annual freight costs of $66,700. The net
transportation savings from the split of Rockford DC volume was approximately $36,100 per
year.
The impact on inventory due to the proposed changes was also positive. Generally, when
adding a new distribution center, like would be required in the Miramar facility, the impact
would be an overall increase in inventory as part locations increase. In this instance there is no
net increase in part locations because the inventory is not retained in the original site. In fact,
relocating the parts to Miramar and Windsor Locks allowed for the benefits of inventory
reduction from the approaches proposed in section 4.3. The savings in annual inventory carrying
costs by reducing the average inventory and safety stocks from having co-located repair and DC
was approximately $99,000.
The final impact on costs was the change to labor across the three sites. Moving 84% of
parts from the Rockford DC would reduce the workforce needs there, but would shift them to the
other locations. Through discussions with management two of the 20 current Rockford staff
would remain, one would be needed in Windsor Locks and 18 (17 moved from Rockford and
one additional) in Miramar. While Windsor Locks has a similar wage rate to Rockford and
would have no net cost change, Miramar has a considerably lower wage rate. The savings from
the lower, fully burdened labor rates including overtime for the seventeen staff being transferred,
less the addition of one more staff resulted in an annual labor savings of $292,000. The inventory
carrying cost reduction of $99,000 per year, transportation savings of $36,100 per year and labor
savings of $292,000 per year resulted in an estimated impact of $427,100 per year.
4.2.2. Supporting Facts
Splitting the Rockford DC volume is a large undertaking with long term impacts to the
Hamilton Sundstrand aftermarket distribution market. Successfully making the transition would
update the distribution network to the current operating environment and have several positive
side effects:
e
Reduces network transportation expenses: Currently Hamilton Sundstrand pays over
$38,000 per year to supply parts to the Miramar repair facility from Rockford. Those
parts have little use in Rockford, just being stored locally there based on historical
operating procedures before being re-routed to a repair facility, OSS location or external
customer. Moving those parts, as well as the other part family location changes, can save
up to $66,700 per year in outbound transportation costs. Although there is an increase in
inbound transportation expenses, the net savings is $36,100 per year.
" Reduces average inventory in Windsor Locks and Miramar repair: With the change
in part locations comes an increase in the co-located DC and repair inventory stocks.
Implementing the strategies discussed in section 4.3 results in reduced average inventory
at both Windsor Locks and Miramar for a total of $99,000 per year.
" Lowers cost to serve: The relocation of the parts to Miramar saves on transportation and
inventory expenses in the network, but it also provides a lower factor cost for a high labor
content operation. Relocating the parts also relocates the associated staff to the lower cost
area, resulting in annual labor savings of $292,000 per year.
4.2.3. Opposing Facts / Alternative Views
There are significant savings and other benefits to relocating parts out of the Rockford
DC and into the two other sites, but there are negative effects as well as a few large assumptions
to moving the parts, including:
e
Assumes space is available in Miramar facility for distribution center: One major
assumption in this proposed recommendation is that there is actually space available in
the existing Miramar facility to support a DC operation. If there is not, leasing space at a
nearby facility would be add costs and may reduce the ability to implement inventory
savings methods.
" Adds a new distribution center: Although there is no increase in inventory cost from
adding the new location from parts it does add management complexity as one more site
to monitor and maintain performance.
" Splits customer and Hamilton Sundstrand shipments: Many of the parts that would be
relocated to Windsor Locks will benefit the network in consolidated shipments to OSS
locations, and benefit customers in ordering parts from multiple product families from
one site. Most of the parts being relocated, however, increase the operational footprint,
adding more sites from which parts can ship from. This impacts both Hamilton
Sundstrand and external customers in potential reduction of consolidated shipments.
" Labor relationships: Moving staff positions from one location to another is a very
delicate subject, especially when headcount is being moved from a union operation to a
non-union operation like in the proposed alternative. Although there are related savings,
there may be significant impacts to labor relationships, morale and existing contract
provisions that outweigh the gains.
4.2.4. Recommendation
With the caveat that enough space is available in the Miramar repair facility to build out a
distribution center, the author strongly recommends splitting a portion of the volume in the
Rockford DC to Miramar and Windsor Locks. The change will provide substantial annual cost
savings and matches the supply chain strategy of the distribution network.
Implementation of the change requires first working with the Miramar repair facility to
identify potential space. Recent lean manufacturing efforts and other refinements at the repair
facility should be able to provide the required space needed for a small foot print distribution
center. Second, the parts targeted for relocation should be reviewed by stakeholders across the
company, suppliers and customers to ensure they would not have other undesirable side effects
of being moved. Finally, if the proposed part relocation is adopted, the parts targeted for Windsor
Locks should be moved first to test the impact on network operations with a small change. Next
a portion of the parts targeted for Miramar, perhaps one product family, should be moved and
serviced from the new DC to slowly scale up the new operation and train a core set of the new
workers. Once Miramar DC operations have stabilized the rest of the part families can be
relocated.
4.3.
Optimize Co-located Domestic Repair Center Inventory
The strategic inventory placement review found that inventory was placed in mostly
optimal places to buffer supply and demand uncertainty, but that there was significant shared
parts inventory in repair centers that are located adjacent to their supplying distribution centers.
The proximity of the supply source and associated short replenishment time should mean
minimal to no stock for the co-located repair locations, which was not the case with nearly $4
million in duplicate inventory. The supply chain planning review found what may be a key
source of the high levels of inventory in generic inventory policies across the repair sites. While
the distribution centers get significant attention and use time phase planning to optimize
inventory, the repair centers largely depend on a min-max inventory system that uses the
provided parameters to achieve a certain service level. Ensuring optimal parameters, stocking
policies and service levels are in place can make a big impact on the inventory performance.
4.3.1. Repair Center Inventory Planning
Hamilton Sundstrand uses a sophisticated inventory management software tool called
Servigistics that continuously monitors repair center inventory levels and determines order
quantities and order levels based on the desired fill rate. The policy is a commonly used
continuous order-point, order-up-to-level known as an (s, S) or min-max policy. This policy
places an order when the inventory position (IP) is at or below the re-order level (s) with an order
sufficient in size (S - IP) to bring the IP back up to the optimal order-up-to-level (S). Figure 19
shows an example of the min-max (s, S) policy shown over time.
S
-nventory Position
Lead
time
0
Time
Figure 19 - Min-Max (s, S) Continuous Review Order Policy
Sourced from (Simchi-Levi, Simchi-Levi, & Kaminsky, 2004)
The order-up-to-level (S) is determined by calculating an optimal order quantity (Q) and
adding it to the re-order level (s + Q). With the usual demand rate of one unit per order, the
replenishment order size is typically close to or equal to the optimal order quantity. The effective
optimal order quantity is determined as the lower quantity from two approaches, either the
economic order quantity (Q*) or the months of supply maximum. Economic order quantity (Q*)
is determined using Equation 1 with the key inputs of demand, order quantity, unit cost and
holding (carrying) cost. Sometimes the economic order quantity can result in an order size that
could sustain demand for a long period of time, which could be problematic with aircraft parts
that can be made obsolete by engineering changes. The maximum months of supply quantity
caps the maximum order size to control for this; its source is the inventory parameters, recapped
in Figure 17.
EOQ (Q*)=
2*A D
v*r
(Equation 1)
Where:
A = Order cost
D = Annual demand
v = Unit cost
r = Carrying / holding cost rate (%)
The re-order level (s) is determined by adding the demand over lead time (XL) and the
safety stock (ss) for a desired service level together. Typically safety stock (ss) is determined
using the normal distribution with a formula similar to Equation 2.
ss = Z x
fL +R x a
(Equation 2)
Where:
Z = safety factor determined from the desired service level
L = Lead time
R = Review period (0 if continuous)
a = Standard deviation of demand
Assumes time intervals (L & R) are independent
The normal distribution is a good representation for demand characteristics on fast moving parts
and is widely known and used in practice. Many of the parts in the distribution centers do have
sufficient volume to match the normal distribution; however the slow moving and intermittent
nature of a large portion of spare parts demand, especially those at repair centers, is more
appropriately represented by a discrete distribution like Poisson.
Silver, Pyke and Peterson (1998) propose that the Poisson distribution is appropriate to
use when the observed standard deviation over lead time (GL) is within ten percent of the square
root of lead time demand (XL), and when XL is below 10 units. This test is both quick and
effective, as the Poisson distribution has a characteristic of equal mean and variance which is
easy to determine from a data set. The parts at the Windsor Locks repair facility, one of the
larger volume co-located repair sites, were examined to determine their fit to the Poisson
distribution based on these tests. A random selection of the parts tested is shown in Figure 20.
Leadtrne
Ldtudn
sr(
den_.
Product A
Product B
Product C
Product D
Product E
Product F
Product G
Product H
Product I
Product J
4.741
2.533
2.325
1.511
3.259
1.067
7.777
1.141
8.889
1.319
2.177
1.592
1.525
1.229
1.805
1.033
2.789
1.068
2.981
1.148
St& Dev.
1.987
1.496
1.551
1.135
2.054
1.064
3.005
0.872
3.227
1.187
Figure 20 - Poisson Test Part Sample
Over 95% of parts had XL below 10 units and over 75% of parts had GL within ten percent of the
square root of XL. The strong supporting evidence through these tests and Servigistics' own
demand distribution identification approach 5 support a Poisson approach to inventory policies.
The use of a Poisson distribution both simplifies and complicates setting inventory levels.
For slow moving items, it is important to be able to deal with discrete units. On the other hand,
discrete mathematics creates problems in implementation (Silver, Pyke, & Petereson, 1998).
With a normal distribution the safety factor (z) is relatively easy to determine given a desired
item fill rate (IFR) using the unit normal loss function (G[z]). As shown in Equation 3, the
5 Servigistics identifies part demand distributions by analyzing the coefficient of variation (CV = std. dev / mean) for
each part. Parts with CV greater than one are attributed a Poisson distribution.
63
normal loss function value can be determined using the desired item fill rate, standard deviation
over lead time (0L) and order quantity (Q). The safety factor can then be found using the normal
loss function value on a common reference table or using functions in Microsoft Excel, plugged
into the safety stock formula of Equation 1 and a resulting re-order level is found.
G[z] = 2 (1 - IFR)
(Equation 3)
GL
With the Poisson distribution, finding the correct re-order level is slightly more involved. Due to
the discrete nature of the distribution there is no quick inverse loss function to reference. Instead
the appropriate re-order level necessary to meet a desired item fill rate must be found by
incrementally adding units and testing the expected units short based on the demand
characteristics. At each incremental re-order level the expected fill rate can be determined using
Equation 4.
IFR
=
(Equation 4)
-A
Where:
X= mean (and variance) of demand
x = re-order level
Calculating the fill rate for a given re-order level is straightforward using Microsoft
Excel's POISSONO formula. However, there is no built-in inverse Poisson formula to quickly
arrive at a re-order level for a desired fill rate. To test alternative inventory policies without
having to iterate through re-order levels on each part manually the author wrote an Excel
function to perform the task efficiently.
The average inventory on hand for a part is half of its cycle stock (Q/2) plus its safety
stock. Both components are heavily impacted by the inventory parameters dictated to the system.
The cycle inventory relies heavily on the provided ordering cost parameter driving the economic
order quantity and the maximum months of supply. The safety stock is a direct result of the
64
service level as well as the replenishment lead time over which demand uncertainty must be met.
The service level is provided by the inventory parameters, and the replenishment lead time is
manually updated in Servigistics. With the right parameters and ASL policies the inventory
levels can be optimized to meet service level needs without driving unnecessary costs.
4.3.2. Reduce Average Inventory
Reducing inventory for the sake of cutting costs is not a good decision if it goes counter
to service level goals. Generally inventory reduction plans are soon followed by a plan to
increase service levels and vice versa in a constant pendulum effect of balancing costs with
service. However, occasionally there are opportunities in locating inefficiencies in the system
where inventory reduction can be achieved without service impacts. The co-located repair center
parts inventory is one of those opportunities.
To determine the ability to reduce the average inventory on hand while maintaining
service levels two areas were addressed: cycle inventory and safety stock. Cycle stock was
analyzed in the domestic repair centers to determine the optimal parameter levels, primarily
ordering cost, to reduce the average order size. Safety stock levels were analyzed to determine
the appropriate levels based on acceptable service levels and actual replenishment times versus
current parameters.
4.3.2.1.
Estimated Impact
For each repair center an analytical model was built that held all of the commercial ASL
parts managed at the facility and their associated planning parameters. The models provided the
ability to input new parameters, specifically order cost, replenishment lead time and desired
service level, and see the impact on the average inventory levels. With each change in input
parameters the parts would have new order quantities calculated using the current approach of
economic order quantity or maximum months on hand, whichever was less. The model would
also dynamically determine the appropriate re-order level based on the InversePoisson function
written by the author given desired service level and lead time demand, a function of
replenishment lead time. In addition to determining the resulting change in average inventory the
model would also predict the increase in replenishments necessary to meet the new parameters.
This was necessary because driving ordering quantities down means increasing frequency of
orders, which increases distribution center fulfillment activity and associated labor costs.
The average cycle stock inventory was optimized using the ordering cost parameter and
the tradeoff between resulting inventory gains and additional picks in the distribution center. The
original ordering cost, for all repair sites, was $15. This number was believed to be too high for
the domestic repair sites as there were no transportation expenses, transactions were electronic
and automated in ordering systems, and labor expense was limited to picking, consolidating and
carting to the stock room in the adjacent building. With this in mind the ordering cost was
progressively lowered to determine the impact on average inventory costs and additional
distribution picks. This process was repeated individually for the Windsor Locks, Phoenix and
Rockford repair centers. Figure 21 shows the tradeoff at Windsor Locks repair between reduction
in average inventory and additional DC picks as the ordering cost was lowered. Through this
process the optimal ordering cost was determined to be $3.00 as this level resulted in a $391,000
reduction in average inventory while only requiring an estimated 23 extra picks per day, a 26%
increase in distribution work related to local repair replenishment, but only a 2-3% increase in
total picks at the local DC. Going beyond $3.00 at Windsor Locks quickly resulted in a spike in
extra distribution picks without significant gains in inventory reduction. Completing the process
at each facility in resulted in similar ordering costs of approximately $3.00 and a grand total of
$512,000 in average inventory reduction across all sites, or approximately $102,000 in annual
carrying costs.
$700,000
-
100
9
$600,000
so
$5M,000
70
-
8$400,000
so 1
t
0$300,000
420
105
$0
____
$100,000
0
Q OP
r6
Ni
r-4
-
0
_
r-f
O
a Ca C3a
O QO
r4
i
A
W~0 r-
in
II
M
4
O
Q
6o
O
a
A
i
A
A
O
O
4v
M miPiO M
C-i
O
4^
IA
O
Q
a CO
aOM
IA# IA
M
V-# o
I^A^ IAn
Order Cost
Reduction inAvg. Inventory Add'i Distribution Picks/Day
Figure 21 - Windsor Locks Ordering Cost Optimization
The safety stock was optimized using the two inputs of target service level and
replenishment lead time. As each part has a unique service level determined by its demand and
cost characteristics, the service levels were adjusted by a ratio of their current service level
premium, the amount above a 50%, zero safety stock level. For instance, if a part had a current
service level of 90% (service premium of 40%), adjusting the service premium ratio from current
(100%) to half (50%) would result in a 70% service level (50% base + [40% premium x 50%
ratio] = 70%). This approach made it easier to scale down each part at the same time instead of
setting new parameters manually for each part. Reducing the service level would creating more
stock outs in the repair center, but with DC stock close by the customer's repair order would not
be dramatically affected. The replenishment lead time was adjusted down from the current level
of four days (in Windsor Locks repair) to zero in one day increments. The actual order
replenishment history was analyzed to validate that the current four day replenishment lead time
was indeed too high and that repair center stock outs were not as critical. Seventy-seven percent
of orders were fulfilled same day and over 90% within two days, as demonstrated in Figure 22.
There are significant opportunities to improve the actual replenishment lead time even further,
but even with this outcome a reduction from the four day parameter in the system seemed
reasonable.
90.0%
77.0%
80.0%
** 70.0% S
-
0.0% -
50.0% 40.0% 30.0% 20.0% 0.%2.2%
1.1%
1.6%
0.0% -
6.7%
0
4
3
2
1
Days Required for Replenishment
>4
Figure 22 - Windsor Locks Repair Replenishment Lead Time from Local DC
Similar to the average inventory reduction approach, each of these parameters were
reduced incrementally to gauge the impact on service performance and inventory reduction. The
resulting inventory reduction for Windsor Locks repair is shown in Figure 23.
T toap)
feplenise..
%of curment serice Level xesntigAverage
0
Premun (FiNRate35s%) Service Level
56.0% $66%,000
0%
54.8% $669,000
10%
59.A% $669,000
20%
64.5% $669,000
30%
40%
50%
60%
70%
30%
90%
100%
1
$668,000
$667,000
4
3
2
$665,000 $630,000 $604,000
$663,000 $618,000 $584,000
$657,000 $612,000 $570,000
$650,000 $597,000 $539,000
$639,000 $585,000 $511,000
$623,000 $567,000 $466,000
$665,000
$663,000
69.3% $669,000 $660,000
74.1% $669,000 $655,000
78.9% $669,000 $645,000 $589,000
33.1% $669,000 $630,000 $571,000
8.6% $669,000 $598,000 $533,000
953A% $669,000 $542,000 $481,000
56.2% $669,000 $425,000 $290,000
$533,000 $421,000
$485,000 $356,000
$417,000 $276,000
$299,000
$106,000
$149,000
$0
Figure 23 - Windsor Locks Repair Safety Stock Optimization Savings
While there are significant savings for more aggressive reductions in replenishment lead time
and service levels, the gain from just reducing system lead time by two days without a reduction
in service level was $290,000 in Windsor Locks, and totaled $362,000 across the network. The
annual carrying cost for this inventory is $72,400.
4.3.2.2.
Supporting Facts
There are several facts supporting the reduction of inventory at the co-located repair
centers including:
e
Significant savings in inventory reduction with minimal impact to service levels:
Between cycle stock average inventory and safety stock the total inventory reduction
would be $874,000 with an annual carrying cost savings of $174,800. These savings can
be realized without substantial risk to service levels.
" Smoothens demand pattern for source distribution center: Reducing the average
order size provides a better demand signal to the distribution center that replenishes the
repair center. Although the distribution centers already incorporate repair demand into
their forecasts and ordering pattern, the larger repair replenishments could trigger stock
outs more frequently than a smaller replenishment size. An analysis of smaller repair
replenishments on sample parts showed critical DC shortages reduce by almost half and
fill rate increase by 1.5%.
" Reduces network excess and obsolescence: The volatile demand patterns on repair parts
often leads to increased inventory that has a high likelihood of becoming excess or
obsolete when the demand patterns inevitably change. Refining the policies to carry a
more appropriate amount of inventory in repair centers reduces the exposure to this risk.
4.3.2.3.
Opposing Facts / Alternative Views
The adjustment in inventory policies leading to the cost savings also carries with it some
drawbacks, including:
e
Additional picks in distribution centers: The reduction in order sizes leads to more
frequent ordering and the associated increase in distribution activity. Although the
increase in total picks related to the proposed alternative only increase approximately 23% that could approach capacity limits for the facility or staffing.
*
Additional complexity in inventory policies: One of the benefits of the current policies
is that they are simple to administrate and communicate. Changing the ordering cost per
site and the replenishment lead time may require changes in the Servigistics platform and
burden on management to maintain additional standards.
4.3.2.4.
Recommendation and Implementation Approach
Given the substantial savings in inventory reduction from the proposed changes and the
positive side effects of smoothed demand at the distribution centers, it is recommended that the
alternatives are implemented. The drawbacks of additional picks in DCs and additional
complexity can be mitigated by a slow implementation of the changes and if needed, minimal
head count increase in the DCs.
The recommended method to implement the proposed changes is to slowly draw down
the inventory parameters over the course of several months. The first parameter to change would
be the ordering cost which should be reduced by $3 increments each month for four months to
gauge the true impact on added picks in the DC. Similarly, the replenishment lead time should be
reduced from the four day current level to two days, by removing one day per month for two
months. This should occur after the ordering cost reduction to minimize potential issues in
identifying causality if problems do arise. The slow adoption will help identify replenishment
issues and could lead to process improvement that could support further reduction in the
parameters. If the replenishment lead times are adequate at the reduced levels it is recommended
that the organization attempt to draw down the replenishment lead time to zero days at a slower
rate to remove the need for local stocking all together.
4.3.3.
Optimize ASL Stocking Policies
Reducing the inventory levels for ASL parts as discussed in the previous section has
substantial gains in inventory reduction. However, it is only optimizing the inventory for parts
that see enough demand to be considered an ASL part at the repair centers. As detailed in the
inventory turns analysis of the inventory planning section, there are many parts that have very
low or no demand that should not be stocked at all. The current ASL policy is too broadly scoped
in its part cost and order event parameters and too conservative in allowing parts to be stocked
within those groups. The proposed alternative for repair center ASL policies is to separate the
cost and event ranges into smaller tiers and determine the optimal trade off in stocking locally
versus sourcing directly from the distribution center (a non-ASL repair part).
4.3.2.1.
Estimated Impact
The optimization of the ASL policies was completed in an iterative process to first
determine the appropriate amount of segments to create in order events and part costs, and then
in locating which of the resulting "buckets" of parts should not be included on the ASL list. The
segmenting process involved aggregating the annual demand for all parts at the repair centers
which are sourced from their local distribution center. Then segment cut offs were created and
adjusted to ensure each "bucket" had enough parts to substantiate the additional complexity of
more segments, but not too many parts in a bucket to reduce the benefits of a more specific ASL
policy.
Events per Year
Part Cost iers
Annual
$0
$3
$10
$25
$75
$3
$10
$25
$75
$200
$200
$500
$1,000
$500
$1,000
$2500
259
180
92
112
77
53
23
14
$2,500$1,oo,000
21
2-3x / year Every 2-3 mos Monthly 1-2x/month
192
175
135
170
94
135
96
124
76
88
86
66
66
62
53
99
60
47
72
69
38
36
42
48
21
39
35
35
18
25
18
24
27
8
14
17
>i/week
weekly
251
215
115
114
90
79
76
54
45
57
46
35
44
28
11
9
3
2
Figure 24 - Revised ASL Policy Segments
The resulting grid of segments, shown in Figure 24, minimized the buckets with more than 250
parts and buckets with fewer than 20 parts. Fewer parts were allowed in the high dollar ranges as
the increased policy complexity was justified by the potential reduction in inventory savings. The
green cells in the grid represent those part buckets that currently qualify for the ASL list,
approximately 78% of all parts with demand.
The next stage of optimizing which buckets of parts should be allowed on ASL policy
involved several steps. For each bucket a tradeoff between holding and ordering costs had to be
found. The ordering cost used in the model was the recommended new value of $3 per order.
That value was used to determine the estimated annual ordering cost on a per part basis should
the part be sourced directly from the distribution center. This resulting cost was then compared
with the average annual holding cost per part if it were included on the ASL list and stocked
locally.
Events per Year
Part Cost rers
$0
$3
$3
$10
1x/year 2-3x / year Every 2-3 mms Monthly 1-2x/month
$0.12
$0.14
$0.25
$0A40
$0.52i
$1.54
$1.64
$1.12
$0.58
$0.47
$0.25
$0.11
$1.081
$0.56
$0.39
$2.30
$1.05
$0.71
$10
$25
$4.75
$3.071
$2.07
$1.25
$25
$75
$12.34
$8.85
$6.86
$3.05
$75
$200
$39.48
$26.25
$11.67
$7.48
$6.34
$500
$1,000
$93.34
$212.57
$75.06
$85.76
$40.01
$54.84
$446.69
$818.72
$192.10
$605.69
$78.39
$22.65
$46.10
$77.56
$11.98
$30.59
$46.77
$294A5
$302.02
$68.16
$200
$500
$2,500
$1,000
$2,500 $1,000,000
Weekly >1/week
$0.04
$0.08
$138
$2.39
_
_
$8.11
$14.461
$28.11
$3.27
$6.31
$11.92
$38.52
$25.37
Figure 25 - Part Buckets Identified for ASL
The holding and ordering costs were calculated for each part bucket. Buckets where the
holding cost was lower than the ordering costs identified them as qualified for ASL. Figure 25
shows the part grid for Windsor Locks repair with the holding cost per part displayed and ASL
qualified buckets highlighted in pink. The result shows several buckets, colored in green, that are
on the ASL list per the current policy, but were not cost justified to be stored locally. It is only a
few buckets of parts, but that accounts for 418 parts (approximately 9.3% of total) and $138,800
in annual inventory savings (reduced holding costs less additional ordering costs) for Windsor
Locks. Across the other domestic repair centers the same buckets were implemented with an
estimated total network annual savings, including Windsor Locks, of $204,800.
4.3.2.2.
Supporting Facts
The benefits of the proposed ASL policy change are similar to the reduction in average
inventory. There are both direct cost savings as well as related positive improvements, including:
Significant low demand inventory in repair facilities: Much of the inventory in the colocated repair sites sees only minimal demand and can easily be served from the nearby
distribution centers. Approximately $204,000 in annual savings can be realized by
adjusting the ASL policies and relocating parts to the distribution centers.
e
Reduces network excess and obsolescence: Over 7,000 parts with inventory in the
Windsor Locks Repair center do not have sufficient demand to be placed on the ASL list.
More than 2,500 of those parts were on the ASL list for less than 60 days before being
removed as demand changed, and account for $1,031,000 in inventory on hand. Changing
the ASL policy won't benefit these parts, but it will prevent a large amount of future parts
from being classified as ASL and building up stock in the repair site.
4.3.2.3.
Opposing Facts / Alternative Views
Changing the ASL policies has much of the same side effects as reducing EOQs in an
effort to reduce average on hand inventory, as the change effectively increases order fulfillment
in the distribution center. The side effects for the ASL policy changes include:
e
Additional picks in distribution centers: Relocating additional groups of parts to be
served from the distribution center increases the order fulfillment activity significantly.
Instead of carrying stock locally and replenishing in larger order quantities the parts must
be sourced as needed, typically in single unit increments.
" Additional complexity in inventory policies: One of the benefits of the current ASL
policies is that they are simple to administer and communicate. Increasing the ASL policy
complexity may require changes in Servigistics to accommodate more business rules and
would require occasional revisiting to determine if they are still optimal for the current
operating environment.
" Increased exposure to stock outs: The repair centers have very quick turnaround
requirements for customer repair work and need parts stocked with high availability to
meet their commitments. Reducing the number of safety stocks by only stocking at the
distribution center increases the likelihood of that a part may not be available in time.
4.3.2.4.
Recommendation
Following the analysis that just a few small refinements in ASL policy could lead to over
$200,000 in annual cost savings, it is recommended that a more detailed policy be put in place.
The short term gains in reducing inventory on active ASL parts will be just a small fraction of
the long term gains in preventing excess and obsolete parts building up in the repair centers.
As with the other recommendations, the proposed method of implementation is to slowly
expand the ASL policy into the more segmented approach. The process should begin with the
higher dollar parts as those have significant gains as well as require more DC stocking for higher
use parts that should provide a reasonable test of the ability to serve common parts directly from
the DC. The remaining tiers of part costs should be implemented in two additional stages, first
with the $500 and above parts tiers followed by the remaining tiers. It might also be appropriate
to test delaying the addition to or removal from the ASL list by a month or more to gauge the
impact on operations and to weather short term demand spikes before committing volume to the
repair center. Co-located repair center parts not on the ASL should be regularly reviewed, aside
from the current quarterly excess and obsolete reviews, to determine additional inefficiencies in
the ASL policy that could benefit inventory performance if resolved.
5.0
Summary
The recommendations proposed in this thesis address inefficiencies in the network design
and inventory planning of the aftermarket distribution network. The estimated combined savings
from closing the Asia distribution center, relocating Rockford DC volume and refining colocated repair inventory policies is over $1 million per year. This significant impact in cost
savings results from changes that maintain high service levels, modernize the distribution
network to current conditions and reduce duplicate inventory. While the changes do optimize in
very explicit areas, they are the result of a broad analysis of the distribution network that
identified those areas specifically. The universal theme is that it is important to first identify the
right problems before attempting to solve them.
5.1
General Takeaways
The various recommended changes proposed in this thesis were for specific challenges
faced by Hamilton Sundstrand's aftermarket network, but neither the challenges nor the solutions
are restricted to their network. The hierarchal approach can be applied to any supply chain
network with similar benefits. The general characteristics of the analysis that could be used in
other organizations include:
e
Ensuring supply chain strategy matches business strategy
"
Refining supply chain practices for current needs
e
Revisiting and validating assumptions
Matching the supply chain strategy with the overall business strategy is a critical process
that can build significant value if done correctly or destroy value if done poorly. In Hamilton
Sundstrand's aftermarket network the business strategy aimed to deliver high customer service in
terms of high availability, quick response and quality parts. An appropriate supply chain strategy
for the business required high service levels, relatively high (but efficient) inventory and rapid
transportation methods. For other business strategies, for instance in low cost wholesalers, this
supply chain would not be appropriate. There is no one-size-fits-all supply chain, they must be
tailor fit for the specific needs of the business.
The practice of matching the supply chain with the business strategy is not a one-time
occurrence. Business needs change significantly over time as products change, competitors enter
or exit the market and customer needs evolve. Trying to force a supply chain built upon
historical needs to meet current conditions will lead to shrinking profits and missed
opportunities. Hamilton Sundstrand has kept pace with their changing needs by adding facilities
in new markets, focusing on new service levels and developing innovative supply chain
relationships like the OSS locations. Periodically taking a step back from the daily operations to
revisit the match of supply chain strategy with current business needs is important to remaining
competitive.
Revisiting the explicit and tacit assumptions made in operating the supply chain is a
critical step in aligning with current business needs and moving closer to the efficient frontier.
Assumptions are present in many systems and analysis like business cases of investment
decisions, inputs in planning systems and cultural operating practices. In the analysis of the
Hamilton Sundstrand aftermarket there were examples of outdate assumptions in business cases
like the Asia distribution center and untested inputs to the planning system like ordering costs
and stocking policies. Typically the models that are used for business decisions are accurate and
perform appropriately; it is the assumptions that power the model that lead to poor outcomes.
Testing the assumptions with sensitivity analysis or revisiting the logic behind them can lead to
very different decisions.
5.2.
Follow-up Analysis
The analysis of the aftermarket network resulted in a limited group of proposed solutions
based on the initial review of network needs and the time available for analysis. With additional
time and effort a few additional topics could be analyzed in more detail to provide additional
improvements in the network. Other topics include revisiting excess and obsolescence inventory
policies, lead time reduction efforts and full consolidation of inventory at co-located repair and
DC sites.
Excess and obsolescence is a challenging issue for aircraft parts providers. The product life
cycles can be very long with high variation in usage patterns. Additionally, design issues with
parts may be addressed with engineering change orders that immediately make previous part
versions obsolete in some cases. Current network stocking policies may have room for
improvement by increasing review frequency, centralizing stocks and adjusting inventory
planning parameters to optimize very slow moving parts without sacrificing service levels
substantially.
As described several times in this thesis, inventory is used to buffer against lead time
differences in supply and demand. With supply lead times far in excess of demand lead times no
amount of inventory optimization is going to rid the network of large stocks of parts. Only
efforts to reduce the supply lead time or increase the demand lead time will dramatically impact
the inventory levels. With high availability goals and stringent customer lead time requirements,
lengthening the demand lead time is probably not appropriate for many customers. That leaves
reducing the supply lead time to impact inventory. Hamilton Sundstrand already has many lead
time reduction efforts in place both with internal OEM suppliers and through supplier
integration. Continuing to seek out methods to reduce the lead times will pay dividends in
reduced inventories by decreasing the demand over lead time component of the safety stock
calculation (see Equation 2 in section 4.3).
Finally, additional gains can be made in co-located repair and DC inventory reduction by
completely consolidating the shared inventories. Although full stock consolidation would not be
possible due to point-of-use inventory and parts sourced directly from other suppliers, the
majority of inventory could be combined into the DC to reduce overall inventory. The pooling of
inventory at a central location saves inventory costs by reducing the required safety stocks due to
lower overall variability in aggregate demand. This effect, called risk pooling, generally reduces
network safety stocks by the square root of the number of locations, but it is dependent on the
underlying variation in demand. With high demand variation, like that experienced in
aftermarket parts, the benefits of inventory pooling can be even greater. Accomplishing complete
consolidation of DC and repair inventories would require changes in daily operating activities
and could face organizational and political barriers to implementation. Gradual shifts in
increased consolidation using more aggressive ASL policies could be used to surface issues and
allow time to adapt for a successful implementation.
Bibliography
UTC 10K Filing.(2009). Retrieved February 8, 2011, from UTC Investor Relations:
http://www.utc.com
Supplier Development. (2010). Retrieved February 8, 2011, from About UTC:
http://investors.utc.com/About+UTC/Suppliers+%26+Partners/Supplier+Development
Andreoli, D., Goodchild, A., & Vitasek, K. (2010). The rise of mega distribution centers and the
impact on logistical uncertainty. TransportationLetters: The InternationalJournalof
TransportationResearch, 75-88.
Beckman, S. L., & Rosenfield, D. B. (2008). OperationsStrategy: Competing in the 21st
Century. New York: McGraw Hill/Irwin.
Byrnes, J. L. (2010). Islands ofProfit in a Sea ofRed Ink. New York: Penguin.
Chopra, S., & Meindl, P. (2007). Supply Chain Management: Strategy, Planningand Operation.
Upper Saddle River, NJ: Pearson Prentice Hall.
Fisher, M. L. (1997). What Is the Right Supply Chain for Your Product? HarvardBusiness
Review, 105-116.
Fortune Magazine. (2010). Fortune 500 Ranking. Retrieved February 8, 2011, from Fortune
Magazine: http://money.cnn.com/magazines/fortune/fortune500/2010/fulllist/
Hamilton Sundstrand Communications. (2010). 2010 Hamilton SundstrandBrochure. Retrieved
February 7, 2011, from
http://www.hamiltonsundstrand.com/StaticFiles/HS/Communications/General/Document
s/HS_2010_brochure.pdf
Hamilton Sundstrand Communications. (2010). About Us. Retrieved February 7, 2011, from
Hamilton Sundstrand.
Hayes, R. H., & Wheelwright, S. C. (1984). Restoring Our Competitive Edge: Competing
Through Manufacturing.New York City: John Wiley.
Silver, E. A., Pyke, D. F., & Petereson, R. (1998). Inventory Management and Production
Planningand Scheduling. New York: John Wiley & Sons, Inc.
Simchi-Levi, D., Simchi-Levi, E., & Kaminsky, P. (2003). Designingand managing the supply
chain: concepts, strategies, and case studies. New York: McGraw Hill/Irwin.
Simchi-Levi, D., Simchi-Levi, E., & Kaminsky, P. (2004). Managingthe supply chain: the
definitive guidefor the business professional.New York: McGraw Hill.
Skinner, W. (1974). The Focused Factory. HarvardBusiness Review, 113-121.